2019
NOTE: This is the raw, unedited, unformatted data and insight going into the 2019 edition, provided here to make it easier to access some of the links. It corresponds somewhat to the print edition. Images may or may not work, etc.
-
2019 Update
This is the complete set of data and articles gathered throughout 2018 and into 2019, for the 2019 edition. The categories include evidence of impact on jobs, interesting advances, opposing viewpoints, etc. If you are reading the print edition, and want clickable links, etc., see: http://tsunami.ai/2019
Disclaimer: the pace of AI advances is so fast, I kept gathering material and pushing back the publication date of the 2019 edition; I finally decided to make it an appendix in the print edition. If you haven’t read the earlier part of the book you may want to start there. My apologies for this appendix being a bit rough around the edges; I had decided it would be better to release than to wait until it is perfect. Feel free to email me feedback or questions at tekelsey@gmail.com - and please feel free to connect with me on LinkedIn: http://linkedin.com/in/tekelsey
AI in Five Minutes: 2018-2019 Milestones
I think this year’s theme could be: AI Affects Everyone.
If you had only five minutes to glance through this, these are some of the things that stuck out to me since writing the first edition of Surfing the Tsunami.
MIT announces Interdisciplinary AI College
The President of MIT’s official release calling on education and government to support AI, as well as interdisciplinary AI education, to help people become “AI Bilingual”:
http://news.mit.edu/2019/president-reif-oped-federal-opportunity-and-threat-ai-0211
Why This Is Important: because major institutions are starting to not only see the importance of AI, but to also see how it’s not just a matter for computer scientists - MIT is seeking to help students in every discipline to learn how to leverage AI. I believe their goal is supported by the data I’ve seen, and that this has implications for individuals and educational institutions.
Google/Amazon release all their internal training material on machine learning for the public.
Learn from ML experts at Google
Whether you’re just learning to code or you’re a seasoned machine learning practitioner, you’ll find information and exercises to help you develop your skills and advance your projects.
Amazon’s own ‘Machine Learning University’ now available to all developers
https://aws.amazon.com/training/learning-paths/machine-learning/
(includes paths for business decision makers)
Plus certification: aws.training/machinelearning
Why This Is Important: Not only are these great resources for learners and educators, but it shows how the need for AI resources for “all” people is becoming increasingly important.
Kai Fu Lee challenges individuals, educators and governments to face AI.
Facial and emotional recognition; how one man is advancing artificial intelligence
(Video - 60 Minutes - 1.13.19)
Scott Pelley reports on the developments in artificial intelligence brought about by venture capitalist Kai-Fu Lee’s investments and China's effort to dominate the AI field.
Why This Is Important: Kai Fu Lee is an AI expert with valuable insight from working in both the United States and China - and he is kind; he cares.
Joseph Aoun, President of Northeastern University, calls for lifelong learning, in light of AI
When it comes to lifelong learning, are universities the providers of last resort?
https://www.linkedin.com/pulse/when-comes-lifelong-learning-universities-providers-last-joseph-aoun/
“It is no longer debatable that artificial intelligence will transform our economy; the only question is the precise size of the tsunami. Numerous studies predict that up to 50 percent of today’s jobs will disappear over the next two decades. In developing economies it could be up to 70 percent.”
> A good article that talks about the implications of AI on the job market as well as the need for lifelong learning for those who either never had college or graduated from college.
> And he even used the word tsunami!
Why This Is Important: more solid evidence of the need for people of all ages to learn more, and for educators to develop life-long strategies.
Individuals and educators take action (This means you, and me)
You: This means you. You’re reading this; you’re taking action. Our time is precious because it is finite. But I think this issue is worth investing time in. I encourage you to invest in your future, the future of your family, and country, by setting aside time to read the entire book, and other books (like Master Algorithm, especially) and every article I mention, and start learning more. Just by starting to read, you’re at the Adapt stage - you’re adapting to AI. I invite you to start moving towards Adopt - learn about the platforms. And in the Learn AI section, you can take steps to become more adept in AI. (What? Adapt/Adopt/Adept? See the Introduction)
Me: Here’s my reaction to the announcement from Rafael Reif that calls for interdisciplinary action and education:
Helping Students to Become AI Bilingual - with Personal Data
https://www.linkedin.com/pulse/helping-students-become-ai-bilingual-personal-data-todd-kelsey/
Why This Is Important: We all need to find ways of helping each other to face AI, for the sake of our careers, our families and countries. On the note of families, the article above is an epiphany I had, realizing that personal data and life stories might be a meaningful way to help people learn more about data, from the vantage point of their own life stories and the ancestry and heritage of their family and community.
For more information about Rafael Reif of MIT, Joseph Aoun, and Kai Fu Lee see the “Learn from Leaders” section for more information and perspective from
Learn from Leaders
I’m highlighting what I think is some of the most important, relevant and insightful coverage on AI in 2018 and 2019, from several leaders who have important perspective on AI. Their perspective has helped inspire me to learn more myself as well as to get the 2019 edition of the book done so I could share it with you.
Rafael Reif, President of MIT
Fact: MIT is developing a new AI college, and seeking to help students in every discipline to increase their technical skills and to leverage AI
Implication: you don’t have to be MIT to come to the same conclusion, and personally I think collaboration is necessary between schools, all their departments, companies and governments around the world to help students in any school to learn more, as well as anyone, of any age, who is not currently in school. That’s the main reason I wrote this book.
The President of MIT’s official release calling on education and government to support AI, as well as interdisciplinary AI education.
http://news.mit.edu/2019/president-reif-oped-federal-opportunity-and-threat-ai-0211
A related related opinion piece he wrote discussing AI bilinguality:
https://www.ft.com/content/24f18c28-2a39-11e9-9222-7024d72222bc
A nice video clip from Switzerland (brrr!) with the same message:
Joseph Aoun, President of Northeastern University
With Changing Students and Times, Colleges Are Going Back to School
https://mobile.nytimes.com/2018/04/05/education/learning/colleges-adapt-changing-students.html
(Comment from Todd: The kind of information pointed to by the excerpt below leads me to believe that facing this issue is an urgent question; the bolded statement (my bold) seems to additional confirmation that this is a now issue, not a future issue.)
—
That keeps Joseph E. Aoun, president of Northeastern University in Boston, up at night. While other presidents in local college towns worry about competing for endowments and enrollment, Mr. Aoun sees another threat: robots.
More than the latest polls, he is driven by a 2013 Oxford University Study that predicted that nearly half of the jobs in the United States are at risk of being taken over by computers within the next two decades. “We said if robots are going to replace human beings we need to help students to be robot-proof, and we built a strategic plan around that,” Mr. Aoun said.
That thinking positioned Mr. Aoun on the fringe of higher education strategizing just a few years ago, but he is now called on weekly to advise other institutions on how to help their students outsmart the workers of the future.
--
When it comes to lifelong learning, are universities the providers of last resort?
https://www.linkedin.com/pulse/when-comes-lifelong-learning-universities-providers-last-joseph-aoun/
> A good article that talks about the implications of AI on the job market as well as the need for lifelong learning for those who either never had college or graduated from college.
He even used the word tsunami!
--
It is no longer debatable that artificial intelligence will transform our economy; the only question is the precise size of the tsunami. Numerous studies predict that up to 50 percent of today’s jobs will disappear over the next two decades. In developing economies it could be up to 70 percent.
These same forces will create new jobs that we can’t even dream of today. But those jobs alone will not be sufficient: Just one in 10 workers are currently employed in fields that are projected to grow. Additionally, more than one in three existing jobs will require entirely different skill sets in the future.
--
Joseph also has a good book that may be of interest to those in higher education.
(There’s even an audiobook version - woohoo!
https://www.amazon.com/gp/aw/d/0262535971/ref=tmm_pap_title_0)
Kai Fu Lee, Expert on AI in U.S., China
Kai Fu Lee is an AI expert and venture capitalist who has concrete advice and perspective about AI and its impact on jobs, as well as the important of AI education, and the need for government support. He is a leader in China but spent many years in the U.S. and has gone out of his way to challenge all governments to face AI (including supporting education and facing the impact of job displacement).
CBS 60 Minutes: Facial and emotional recognition; how one man is advancing artificial intelligence
(Video - 1.13.19)
Scott Pelley reports on the developments in artificial intelligence brought about by venture capitalist Kai-Fu Lee’s investments and China's effort to dominate the AI field
How to Prepare for AI Job Displacement
https://www.linkedin.com/pulse/how-prepare-ai-job-displacement-kai-fu-lee/
As individuals, we should accept that routine jobs are going away. For young people in these routine jobs, start now by finding careers that fit your strengths and that are not easily replaced by AI. For older people, when early retirement is offered to you, consider accepting, with gig economy and volunteering to make some income and live a life you enjoy.
We should encourage more people to go into service careers, choosing jobs into which they can pour their hearts and souls, spreading their love and experiences.
We should embrace AI tools, especially for professionals, understanding that they will get better with more data and use. We should use these tools to do parts of our jobs, allowing them to do more of our routine tasks, freeing us to move into areas that are more suitable for humans.
The Human Promise of the AI Revolution
Artificial intelligence will radically disrupt the world of work, but the right policy choices can make it a force for a more compassionate social contract.
https://www.wsj.com/articles/the-human-promise-of-the-ai-revolution-1536935115?mod=mhp
The new technology will wipe out a huge portion of work as we’ve known it, dramatically widening the wealth gap and posing a challenge to the human dignity of us all.
This unprecedented disruption requires no new scientific breakthroughs in AI, just the application of existing technology to new problems. It will hit many white-collar professionals just as hard as it hits blue-collar factory workers.
—
According to a June 2017 study by the consulting firm PwC, AI’s advance will generate $15.7 trillion in additional wealth for the world by 2030. This is great news for those with access to large amounts of capital and data. It’s very bad news for anyone who earns their living doing soon-to-be-replaced jobs.
—
The jobs that will remain relatively insulated from AI fall on opposite ends of the income spectrum. CEOs, home care nurses, attorneys and hairstylists are all in “safe” professions, but the people in some of these professions will be swimming in the riches of the AI revolution while others compete against a vast pool of desperate fellow workers.
—
BIO
Kai-Fu Lee is a Chinese venture capitalist, technology executive, writer, and an artificial intelligence (AI) expert. He is currently based in Beijing, China. In his book, published in 2018, AI Superpowers: China, Silicon Valley, and the New World Order Lee described how China was rapidly moving forward to become the global leader in AI, and may well surpass the United States, because of China's demographics and its amassing of huge data sets.
In a 28 September 2018 interview on the PBS Amanpour program, he emphasized that AI, with all its capabilities, will never be capable of creativity or empathy.
https://en.wikipedia.org/wiki/Kai-Fu_Lee
Kai’s New Book: AI Superpowers
I think Kai Fu Lee’s book is worth reading, to learn more about how AI works, the related issues, and opportunities for the future for individuals and countries. In the presentation I attended in Chicago in September of 2019, Kai explained that the publisher pushed for the stark national imagery on the cover. But as a person Kai Fu Lee is mild-mannered, and his perspective is very valuable. The book does touch on international themes - but the conclusions are not really “us vs them” - one of the most interesting conclusions is the interesting idea that governments can and should encourage the development of jobs requiring compassion and empathy. (See the CBS 60 Minutes video mentioned above)
https://www.amazon.com/AI-Superpowers-China-Silicon-Valley/dp/0358105587
Note: it’s the #1 release in Government Management, and #1 in other categories.
My Thoughts: Kai Fu Lee is cool. I like how he is a superhero venture capitalist in China, but also had deep experience in the USA - but through life experience he is also compassionate, and calls on governments to support education, and the creation of jobs which AI may not be able to overcome (which to him, astonishingly and interestingly, means jobs involving compassion and love - he already assumes and cites evidence showing how AI will take many, many jobs away, even as it creates others)
I had the chance to meet Kai Fu Lee in September of 2018, and I had just read a good blog article that talked about a delineation emerging in data science jobs of data analyst, data scientist, and data engineers. Data analysts probably analyze, report, maybe visualize data. Data scientists work more deeply with the technical tools, and manage things like machine learning, neural networks and deep learning. Data Engineers would be the people who learn how data works, where it resides in companies, how to connect the systems, and the vast amount of work that goes into preparing data in order for AI and machine learning to be able to work with it.
So in the presentation he gave, which was on his book AI Superpowers (see bio below), I asked him, if we consider that AI has increasing powers to automate just about anything, what about automation of AI jobs themselves? I realized and proposed that maybe data engineers might have the most job security long term, because aspects of data analysis and data science might be easier to automate (not easy, but as AI increases in sophistication, witnessed even in the present by things like Google’s AutoML, which was driven by the AI talent shortage to become an attempt to have AI create and manage itself, reducing the need for human input).
Kai agreed that data engineering might be the most secure, and maybe where the most positions will be. (For those interested in Enterprise AI and data engineering, see http://tsunami.ai/white-papers and look for the data integration paper from Amazon.)
Kai was kind enough to let me take a picture; and also kind enough to do an interview for the 2019 edition of Surfing the Tsunami.
My Favorite Article of 2018-2019
One of the main reasons I wrote this book is because I became convinced over time to take AI very seriously, and to explore the data to see both the opportunities and disruption of AI. To me the data points to more jobs being lost than created, but it is a complex issue and estimates do vary. To me, the very fact that advances in AI are so unpredictable, is one of the top reasons why I am skeptical about any claim that there will be more jobs created. The next reason is because it seems that so much of work is already digital; that advances in AI could automate work significantly faster than jobs can be created. My response has been to try and help people learn as much as possible, and to think about methods of creating jobs with AI.
This past year, I came across a third reason. In the following article, they talk about how 90% of the work in AI is working with data preparation (which bodes well for data engineers). So I might call this data inertia, the barriers to AI implementation. And on the surface it would seem like this is evidence of AI “slowing down”, or less chance of it taking jobs.
Net Job Loss/debate - #1 article - 90% rule
When Genpact, an IT services company, helps businesses launch what they consider AI projects, “10% of the work is AI,” says Sanjay Srivastava, the chief digital officer. “Ninety percent of the work is actually data extraction, cleansing, normalizing, wrangling.”
—
https://www.technologyreview.com/s/612897/this-is-why-ai-has-yet-to-reshape-most-businesses/
Inspiration: Applications of AI
Finance
How machine learning and data science give Bloomberg a competitive advantage (Video, article)
CTO Shawn Edwards says strong AI capability helps the firm build data-led products for its customers.
> This also is an example of how AI = data science = machine learning
Business Intelligence in Finance – Current Applications (Updated 2.19)
https://www.techemergence.com/business-intelligence-in-finance-current-applications/
Reuters referenced a Stratistics MRC figure estimating the size of the business intelligence industry around $15.64 billion in 2016. It follows that AI would find its way into the business intelligence world. In our previous report, we covered 6 use-cases for AI in business intelligence. As of now, numerous companies claim to assist business leaders in the finance domain, specifically, in aspects of their roles using AI. We researched the space to better understand where AI comes into play in business intelligence in the finance industry and to answer the following questions
> Tools mentioned include RapidMiner, DataRobot, Domo
> TechEmergence, now Emerj, has a lot of good resources on the emerging AI industry and resources for connecting with it.
Wall Street Tech Spree: With Kensho Acquisition S&P Global Makes Largest A.I. Deal In History
Nadler’s initial plan for Kensho: Use machine learning to make complex financial analysis as easy as a search on Google.
Small Business
https://www.techemergence.com/how-machine-learning-will-become-accessible-to-small-businesses/
Real Estate
https://www.wsj.com/articles/how-to-buy-a-house-the-wall-street-way-1537102800?mod=mhp
> an article about how AI was applied in the real estate sector, for competitive advantage
Marketing
Assessing the Intelligence of AI Marketing Tech (2.19)
General
7 popular AI use cases today
Implementing AI
Data Management Experts Share Best Practices for Machine Learning (2.19)
AI for Business
AI Companies
For AI companies, see below in white papers or http://tsunami.ai/white-papers - and look for the list of top 100 AI companies. And of course, google for “top AI companies _____” (whatever year it is) and “top AI startups ______” (ex: 2019, 2020, etc.)
Also: https://www.cbinsights.com/research/artificial-intelligence-top-startups/
Emerging/Related Technology
Driverless Cars
Why you have (probably) already bought your last car
https://www.bbc.com/news/business-45786690
Drones
drones + planes = bad
https://m.youtube.com/watch?v=QH0V7kp-xg0
Inspiration: Opportunities (and Jobs) in AI
AI and Science
A new approach to detecting cancer earlier from blood tests
https://www.sciencedaily.com/releases/2018/11/181114132000.htm
Cancer scientists have combined 'liquid biopsy,' epigenetic alterations and machine learning to develop a blood test to detect and classify cancer at its earliest stages.
Note: Science Daily is an awesome site. I recommend bookmarking it!
A.I. Is Helping Scientists Predict When and Where the Next Big Earthquake Will Be
https://www.nytimes.com/2018/10/26/technology/earthquake-predictions-artificial-intelligence.html
A nice article showing how AI-assisted robotics could help with prosthetic limbs
https://www.nytimes.com/interactive/2018/07/30/technology/robot-hands.html?
This is VR, but it is still cool. I’m sure there will be a bit of AI in their somewhere.
Jobs/Careers in AI/Data Science/Machine Learning
Jobs in Data Science: data analysts, data scientists, data engineers
https://www.datacamp.com/community/blog/data-engineering
> Important discussion to help you understand the kinds of jobs that exist in AI. Keep in mind that this will change over time. At the present time, jobs in AI are dominated by roles requiring a “theoretical” understanding of AI: all the math, coding and other foundations; but over time as data science and AI flow from academic programs into applied use in business, there will be more “applied” roles. The article above talks about the industry using the language of data science, but other related job titles include Machine Learning Engineer. To help understand the kind of roles available, I invite you to go to indeed.com and jobs on Linkedin, and search for various keywords, including the titles above. You might also want to look at salaries in indeed as well as on salary.com
Future-proof your IT career with these critical skills
In the era of digital transformation, IT pros must adjust to a rapidly shifting technology and business landscape. Here’s a long-term look at where to aim your career in the years ahead.
Demand for AI talent exploding: Here are the 10 most in-demand jobs
Employer demand for AI-related roles has more than doubled over the past three years, according to Indeed.
At the present time, and probably for the near to mid-term, there is a severe shortage of AI, deep learning and neural network talent. To give you some idea of the consequences:
How much should I charge per hour for a deep learning consulting job?
https://www.quora.com/How-much-should-I-charge-per-hour-for-a-deep-learning-consulting-job
Careers in Data Science
Understanding Data Science and Why It’s So Important
It’s been said that Data Scientist is the“sexiest job title of the 21st century.” Why is it such a demanded position these days? The short answer is that over the last decade there’s been a massive explosion in both the data generated and retained by companies, as well as you and me. Sometimes we call this “big data,” and like a pile of lumber we’d like to build something with it. Data scientists are the people who make sense out of all this data and figure out just what can be done with it.
My Journey Into Data Science
https://towardsdatascience.com/my-journey-into-data-science-39e9bbbbf452
Day in the Life: Data Scientist
https://www.youtube.com/watch?v=_Wk9T_G-u4o
In this episode of our “Day in the Life” series, Chevron Data Scientist Alena Crivello describes her use of analytics to find solutions to complex problems.
Career Transition Towards Data Analytics & Science. Here’s my Story
The Role of Statistics in Data Science
“ . . . statistics is foundational to data science—along with database management and distributed and parallel systems—and its use in this emerging field empowers researchers to extract knowledge and obtain better results from big Data and other analytics projects. The statement also encourages maximum and multifaceted collaboration between statisticians and data scientists to maximize the full potential of data science.”
AI For Social Good
Really good article on using AI for good; may be inspiring to workers, students, educators and provide some ideas for areas to pursue, or futures to create.
https://utterbuzz.com/2018/10/ai-for-social-good/
AI Optimism/Job Creation
There are definitely people who believe that AI will result in net job creation, and that AI is pretty much for the better. To me the book Master Algorithm is a key example - that book inspired me when I was alarmed, and helped me cross over into wanting to learn AI.
Google DeepMind founder Demis Hassabis: Three truths about AI
https://www.techrepublic.com/article/google-deepmind-founder-demis-hassabis-three-truths-about-ai/
One of the creators of the AI research company famed for building the pioneering AlphaGo AI spells out the technology's impact and future development.
The New Form of Intelligence You'll Need
Artificial intelligence and human IQ might be what schools and employers will be looking for in the future.
From rust belt to robot belt: Turning AI into jobs in the US heartland
https://www.technologyreview.com/s/611412/ai-could-wreak-economic-havoc-we-need-more-of-it/
Artificial intelligence is offering an amazing opportunity to increase prosperity, but whether or not we will seize it is our choice.
https://www.wsj.com/articles/seven-jobs-robots-will-createor-expand-1525054021
An optimist view of net job growth for AI, and expected in-demand jobs
http://techgenix.com/in-demand-it-positions/amp/
Optimist Viewpoint on WEF Study
Most of the studies I’ve seen indicate there will be significant job loss from AI - but in some cases the same study will say different things to different people. I also believe it’s important to consider opposing viewpoints.
Here’s an example of an optimist blurb from a World Economic Forum study:
https://247wallst.com/economy/2018/09/17/rise-of-machines-and-ai-could-add-133-million-jobs/
Job Creation
Learn AI
60 Second Suggestion: download datacamp, codeacademy, mimo, and check out the free Google courses.
My general suggestion is to explore and try. Read through some of the options below, and try some.
I think it’s worth watching videos and reading about machine learning (including applications of AI, mentioned in the book), deep learning, even if you aren’t “ready”. As mentioned in the review of Master Algorithm - that got me inspired to want to learn how to actually do it. When you start wanting to get into it directly, there are some pre-requisites, and if you reach that milestone, I still recommend making a sandwich out of it - continuing to learn about the high level (deep learning, machine learning), and learning the pre-requisites in parallel: coding, linear algebra, basic probability, statistics). Ideally, when learning the pre-requisites, trying to find material that is focused on the context of machine learning - that is, not just the “entirety” of linear algebra for example (because you might feel overwhelmed); but when possible, looking for material that helps you learn math “for” machine learning, to make it more focused. I think it is the same for coding: even though general introductions are good, when possible try to find material on “coding for machine learning” or “python for data science” or “python for deep learning”, and if you get in over your head, file it away, learn, and come back for more.
Apps: For me, a starting point has just been to download some of the apps, like datacamp, codeacademy, mimo - and just spend a few minutes exploring after dinner.
Games: Because I have to brush up on my linear algebra, I found the very popular app Dragonbox. I’m trying to find more games that teach the relevant topics in AI (please let me know if you know of any - http://tsunami.ai/contact)
Videos: I’ve also found videos a nice way to learn - the 3Blue1Brown videos on linear algebra and the series on deep learning are very popular, and helpful.
Buddies: When possible, find others to go through the journey with you.
Meetups: Meetups are a great way to learn and to find inspiration and encouragement (as well as to ask questions and even find jobs)
Classes: when possible, I think in-person classes are best. If your school or community college or college doesn’t have any - ask them to! I think in-person classes are the best way to learn because of the structure, the human dimension, the ability to ask questions, and the direct mentorship.
Online classes: I think online classes are a great way to learn, to supplement in-person classes, or when they aren’t feasible. In addition to teaching in-person classes, I teach (and take) online classes.
Post/Blog/Share: when possible, I recommend sharing how things are going for you on social media. Don’t worry about being an expert. You may be surprised by the encouragement you’ll find when you share, on personal networks and messaging apps like Facebook, Twitter, Instagram, Pinterest, whatever - or more professional networks. I’m guessing you’ll also find other people who want to learn.
Google AI Education (free)
Learn from ML experts at Google
Whether you’re just learning to code or you’re a seasoned machine learning practitioner, you’ll find information and exercises to help you develop your skills and advance your projects.
Amazon (free, and paid)
Amazon’s own ‘Machine Learning University’ now available to all developers
https://aws.amazon.com/training/learning-paths/machine-learning/
(includes paths for business decision makers)
Plus certification: aws.training/machinelearning
Kaggle
Kaggle.com is a good all around resource. There are learning tutorials, and there are also live projects, competitions, and a sense of community. One prominent data scientists told me that he highly recommended kaggle.
AI for Everyone - AI Expert Andrew Ng - (paid)
Andrew Ng launches ‘AI for Everyone,’ a new Coursera program aimed at business professionals
> Coursera has an audit option for many but not all of their courses
General Topics
Machine Learning
Machine Learning Basics
https://www.datacamp.com/community/news/machine-learning-basics-a9h0fv91cqs
Trading platform example (using machine learning to automate trading)
Introductions to Deep Learning, Neural Networks
But what *is* a Neural Network? | Deep learning, chapter 1
https://www.youtube.com/watch?v=aircAruvnKk&t=468s
> in the Show More/Description area there is more information and a link to a free book he recommends reading.
> Also in the series of videos he recommends a number of resources
Note: this same author has a really good series on linear algebra, which is one of the pre-requisites for the current generation of machine learning.
Some Beginner Intros and Series:
Beginner Intro to Neural Networks 1: Data and Graphing
https://www.youtube.com/watch?v=ZzWaow1Rvho&list=PLxt59R_fWVzT9bDxA76AHm3ig0Gg9S3So
A friendly introduction to Recurrent Neural Networks
https://www.youtube.com/watch?v=UNmqTiOnRfg
Math for Machine Learning
As assistive platforms arise, there may be less need for a better understanding of math, but it is required for the current generation of machine learning and AI. The general types of math related to deep learning and machine learning are basic probability and statistics, linear algebra.
Math in Data Science
https://www.dataquest.io/blog/math-in-data-science/
Learn Math for machine learning.
https://www.kaggle.com/getting-started/59541
--
blogs
https://towardsdatascience.com/the-mathematics-of-machine-learning-894f046c568
https://blog.ycombinator.com/learning-math-for-machine-learning/
https://medium.com/technomancy/the-math-required-for-machine-learning-af0d90db3903
--
videos
https://www.youtube.com/watch?v=8onB7rPG4Pk&vl=en
--
https://courses.washington.edu/css490/2012.Winter/lecture_slides/02_math_essentials.pdf
--
intermediate
https://www.edx.org/course/essential-math-for-machine-learning-python-edition
--
live online course ($1,250 USD)
https://www.thisismetis.com/courses/beginner-python-and-math-for-data-science#overview
Codeacademy
https://www.codecademy.com/catalog/subject/all
Learning coding is closely tied to AI, data science and machine learning. DataCamp, Coursera and Udacity all have intro courses for programming - and Codeacademy is a primary resource that many use, for free and paid coding courses.
Like other providers, there are nice apps you can download, as well as online options.
Below are some of the options
As you grow in your learning, Codeacademy (and Coursera/Udacity) have more advanced and intensive courses. If you are working somewhere, you might be able to get some assistance from your employer. Want some help proposing that? http://tsunami.ai/contact
DataCamp App/Courses
I like Datacamp. Datacamp has a really good intro free app, and good free online courses.
An email below from Nov 2018, some of the promoted courses are intermediate level but they also have intros.
Analyzing Social Media sounds fun, and maybe inspiring to students. Basing data science education on Google Sheets is nice since it is free. I think that starting at the spreadsheet level for data science makes a lot of sense, because it is an easier platform, and may already be familiar to some people. (Google Sheets is a free online alternative to Microsoft Excel - you can access it free and many other tools by starting a google account/gmail account.)
We’re launching eight new courses, including two new spreadsheet courses. Intermediate Spreadsheets for Data Science will expand your Google Sheets vocabulary. You'll dive deep into data types, practice manipulating numeric and logical data, and much more. We are also launching Pivot Tables with Spreadsheets. In this course, you’ll explore the world of Pivot Tables within Google Sheets, and learn how to quickly organize thousands of data points with just a few clicks of the mouse.
Finally, check out five courses that are top rated by other DataCamp users.
New Courses:
• Intermediate Spreadsheets for Data Science
• Pivot Tables with Spreadsheets
• Visualizing Geospatial Data in Python
• Dealing with Missing Data in R
• Designing and Analyzing Clinical Trials in R
• Financial Analytics in R
• Foundation of Functional Programming with purrr (in R)
• Network Science: A Tidy Approach (in R)
5 Courses Recommended by Other DataCamp Users
• Multiple and Logistic Regression (in R)
• Introduction to Time Series Analysis in Python
• Machine Learning Toolbox (in R)
• Text Mining: Bag of Words (in R)
• Analyzing Social Media Data in Python
See: https://www.datacamp.com/courses
Coursera - free or paid (also look up Udacity)
Coursera has an audit option for many but not all of their courses. Both Coursera and Udacity have a variety of courses, certificates and even degrees available in relation to AI and machine learning. (ex: sponsored by AT&T and given through Georgia Tech and Udacity, there is a Master’s Degree in Computer Science with a Specialization in Machine Learning, which is significantly less expensive than other Master’s Degrees.)
Re: auditing (i.e. free)
https://learner.coursera.help/hc/en-us/articles/209818613-Enrollment-options
https://www.coursera.org/specializations/mathematics-machine-learning
https://www.coursera.org/learn/linear-algebra-machine-learning
Mimo
Mimo is an app for learning coding. I’m impressed.
Mimo has a nice format
And good feedback
It is interactive and uses familiar items to build coding concepts.
Legos are an excellent metaphor. It is said that programmers sometimes actually find them helpful in thinking about modularity
Data Science
Data scientists weigh in: 5 data science tools to consider (August 2018)
The surge of data captured by today's organizations requires data science tools to fully understand the information. We asked data scientists what tools they're using.
LinkedIN (free trial, subscription)
https://www.linkedin.com/learning/python-for-data-science-essential-training
> and many other courses
AI Realism: Understand AI’s Impact on Jobs
The June 2018 Financial Times article on Citi seems to me like one of the most important articles in 2018 in regards to job impact - it made one of the biggest impressions on me, serving as evidence that AI-driven automation is already having a significant impact on jobs.
Finance
June 2018 report that Citi expects to shed 20,000 jobs
Citi issues stark warning on automation of bank jobs | Financial Times
https://www.ft.com/content/579c977c-6d73-11e8-92d3-6c13e5c92914
Jun 11, 2018 - Citigroup's investment bank has suggested that it will shed up to half of its 20,000 technology and operations staff in the next five years
Personal
How to Survive AI
http://cmr.berkeley.edu/blog/2018/10/how-to-survive-ai/
F.I.R.E.: Financial Independence, Retire Early
There is a movement among some people who can do it, to try and be frugal, earn as much as possible and seek to retire early.
https://www.nytimes.com/2018/09/01/style/fire-financial-independence-retire-early.html
This is the paragraph that caught my eye. So these are well-paid tech sector people we’re talking about.
Though they had good educations and well-paying jobs in the booming tech sector, Ms. Shen and Mr. Leung faced the looming threats of outsourcing and artificial intelligence, and had no hope of a retirement pension, or even that their employers would exist in five years.
Retail
Good video
https://www.cbinsights.com/research/cashierless-retail-technologies-companies-trends/
Healthcare
Google’s AI is better at spotting advanced breast cancer than pathologists
The firm’s deep-learning tool was able to correctly distinguish metastatic cancer 99% of the time, a greater accuracy rate than human pathologists.
AI is making significant inroads into health. On the good side, there are many medical discoveries waiting to happen, and already happening, because of the significant power AI has to process medical data, scientific data, and help build solutions. It is one reason there was a big controversy over Google seeking to obtain and develop a massive database of cancer related information - then the hospital in question backed out and tried to develop it themselves.
Sloan Kettering’s Cozy Deal With Start-Up Ignites a New Uproar
https://www.nytimes.com/2018/09/20/health/memorial-sloan-kettering-cancer-paige-ai.html
I’m pretty confident medical research is one of the biggest opportunites for AI - could it help cure cancer? Or _______?
And on the other hand, the power of AI will probably reduce the need for medical staff. AI has already proven to be more effective than doctors at diagnosis, because it can compare symptoms to a massive medical database. It is true there will likely always be a need for a human element, and AI Expert Kai Fu Lee recognizes the massive job loss will bring in all fields, and has an interesting perspective - he believes in encourage the growth and creation of jobs involving empathy - including health caregiving. So there is some opportunity.
It will be an ongoing debate.
Automation/Manufacturing
Automation is not always a clear-cut issue. Labor shortage can lead to automation. (But that’s still jobs that could have been done by humans)
https://www.wsj.com/articles/companies-ramp-up-worker-retraining-efforts-1535889600?mod=mhp
The realities of what some people are facing:
Why robots helped Donald Trump win
Toledo has more robots per worker than any other US city. They’re producing a healthy economy—and lots of anxiety.
https://www.technologyreview.com/s/611422/why-robots-helped-donald-trump-win/
—
Some Toledo-area leaders might realize that a technological meteor is headed their way. But what are they supposed to do about something so unpredictable in its details? So they stress “lifelong learning.” From the junior high kids in the robotics camp to the factory-floor employee, they all have to turn their lives into one long hustle to keep their heads above the incoming tide and whatever it might wash in.
—
“We used to laugh at the robots,” Rickey’s buddy said. “When they first came in, they were so slow. We would sorta hurry and outproduce them. But one of the lines was about 18 people, and now they can run it with, like, five.”
—
Rickey looked at me and said he tells his own children that if they wind up working in the plant, “then I failed as a father.”
New autonomous farm wants to produce food without human workers
Down on a new robot farm, machines tend rows of leafy greens under the watch of software called “The Brain.”
Digital Marketing
The original reason why I wrote Surfing the Tsunami is because of teaching digital marketing and starting to wonder throughout 2017 if, or when, digital marketing jobs might be replaced with AI. That led to everything else, including me taking my own medicine, and learning more about data science, AI, linear algebra, Python, machine learning, etc. Frederick Vallaeys is the CEO of Optmyzr.com, whose articles helped inform the beginning of my journey; and the articles continued.
Are the Days of Human-Managed PPC Numbered?
https://www.searchenginejournal.com/automation-human-managed-ppc/252255/
Digital Marketing: The Vallaeys Scale (my term)
Frederick’s comments are about digital marketing specifically, but similar evolution and automation will be affecting every other field.
Todd: So I want to call your automation scale the Vallaeys Scale, kind of like the Richter Scale. Care to give me an estimate of the number we were are at based on dates?
Jan 2017
https://searchengineland.com/artificial-intelligence-drives-ppc-automation-267561
3.0?
May 2018
https://www.searchenginejournal.com/automation-human-managed-ppc/252255
3.5?
August 18
3.6?
AutoML feels to me like the biggest wildcard.
Fred: Nice, I hope people can spell that :-)
I think around 2017 we were at a level 2 because advertisers had to choose from a slew of automation components for each campaign. It did not appear that the automations talked to each other though they did react to what was happening as a result of other automations.
Now (Aug 18) we're at a level 3 where 'smart campaigns' from Google handle lots of aspects in an interconnected way. Smart shopping campaigns may indeed be at a 3.5 because thanks to existing structured data, Google basically knows what is being sold and generates even the ads so there's almost nothing that a human has to do (though there is still lots they can do to improve results)
I don't know I'd put May 18 in a separate bucket. I think a lot of the automations from Google then were still level-2 point solutions.
AutoML is scary interesting though. Could be a game changer!
Talk soon!
Fred Vallaeys
Cofounding CEO
Automation of Writing
Computer Stories: A.I. Is Beginning to Assist Novelists
More on AI’s Impact on Jobs
Watch Out Workers, Algorithms Are Coming to Replace You — Maybe
https://www.nytimes.com/2018/10/18/business/q-and-a-yuval-harari.html
Adapt or die: How to cope when the bots take your job
http://www.bbc.com/news/business-43259906
As the company builds up the number of tasks it can automate, it’s slowly creating a computerized employee.
AI Realism: Understand RPA/Bots
I think robotic process automation is both assisting and destroying many jobs; my tentative view is that RPA is inevitable and inexorable. Attempts to regulate it would probably fail, because other countries would not regulate it and have an economic advantage, and protections would probably eventually be limited or reduced, either by pressure, or companies moving outside the sphere of regulation, or the stagnation of an economy that tried to resist AI. I don’t claim to have the answers, nor am I hostile to RPA. I think I might feel queasy working in a consultancy that was helping to implement it, if a significant number of jobs were to be lost - yet I think it is inevitable, and I’m considering inviting my students to get certified in WorkFusion, for example. Workfusion is mentioned in the alarmist book Rise of The Robots, where it talks about how the robotic process automation engines can be trained by Ivy League grads in sophisticated tasks, to the point where their input is no longer needed. Yet it is also true that automation technology can be assistive - but it still seems to exert considerable force in a likely shrinking of the number of people required to do a particular job, even if a company expands.
So I’d recommend looking at it closely. Robotic Process Automation for me would fit into the Adopt stage of Adapt > Adopt > Adept, ideally where students, workers might have a chance to become the ones who would be managing a platform rather than being replaced by it.
The Big RPA Bubble
https://www.forbes.com/sites/cognitiveworld/2018/12/02/the-big-rpa-bubble/
> This is a really interesting article - it’s aimed at the people implementing robotic process automation, and it seems to be a foregone conclusion that jobs will be replaced. You can infer a lot simply by reading the article.
"What’s the right way to implement RPA? Instead of trying to build a program around staff reduction and thinking about how to achieve 100 percent automation to replace a handful of FTEs, focus on some level of basic task automation for every staff member . . . ”
“Robotic Process Automation (RPA). It’s a hot topic among the C-Suite. Where to implement. How to implement. How many headcounts can be saved through robotic implementations?”
Machine Learning and RPA (Video)
> from a promo email. Video link here: https://tinyurl.com/rpa2-vid or below.
Super smart analysts are the closest things we have to crystal balls in the business world, and Forrester’s Craig Le Clair is one of the brightest crystal balls in Intelligent Automation.
In this free, on-demand webinar, RPA 2.0: the new operation model for scaling and the critical role of analytics, Craig explains:
How the next generation of RPA leverages machine learning (ML) to scale automation
How the future business landscape will be impacted by ML-powered automation
Why it’s critical for enterprises to start their transformation initiatives now
RPA Value Proposition from a promotional email
Our CEO, Alex Lyashok, just announced the launch of Lumen, WorkFusion’s Intelligent Automation 2018 release, which delivers major upgrades to our flagship AI-driven RPA product, Smart Process Automation (SPA), and starter product RPA Express.
The Lumen release aims to make AI simple and powerful for everyone through four core themes:
Simpler RPA: One-click install and faster bot performance.
Everyday AI: Out-of-the-box machine learning for business people.
Enterprise-grade: Centralized and secure credential management for IT compliance.
Analytics: Deeper insights that predict cost, quality, and productivity.
I’d suggest playing with WorkFusion, including RPA Express, in order to understand it better, including perhaps going through the exercise of manually doing a particular task, timing yourself, and then using RPA express to automate the task.
Automation is definitely a thing. There’s even an Automation Anywhere bot Hackathon. And some pitfalls and things to keep in mind when developing and deploying them.
https://blogs.wsj.com/cio/2018/05/29/bots-can-break-leaving-corporate-tasks-undone/
Here is another Workfusion marketing excerpt. This is the kind of material enterprise management is reviewing in consideration of Robotic Process Automation.
--
Hi Todd, Everest Group consistently delivers excellent research on enterprise technology to help businesses adapt and transform. Their latest report, Creating Business Value Through Next-Generation Digital Workforce, is no exception, and we’ve licensed it so that you can freely download it.
You’ll learn the following:
The problems with the current workforce model
The different types of bots that comprise a digital workforce
Why the combination of RPA and cognitive automation is superior to either on its own
--
Why AI Is Unpredictable: Quantum AI
Quantum AI is not the only reason advances in AI are unpredictable, but long-term, it’s a major issue. AI and machine learning could become more sophisticated as a result of quantum computing. That is partly why my view is that automation powered by AI (and/or quantum computing) could replace jobs much more quickly than anyone predicts. Experts in AI, even optimists, usually are careful to qualify their predictions, exactly because of wildcards like quantum computing.
So get in touch with your inner science, and and keep an eye on quantum!
Case in point - an area to watch - machine learning and quantum computing:
Machine learning, meet quantum computing
https://www.technologyreview.com/s/612435/machine-learning-meet-quantum-computing/
A quantum version of the building block behind neural networks could be exponentially more powerful.
Why Most of Us Fail to Grasp Coming Exponential Gains in AI
https://singularityhub.com/2018/07/15/why-most-of-us-fail-to-grasp-coming-exponential-gains-in-ai
> A really good article that shows the unpredictability of the advances (which is why I think it is important to take AI so seriously.
Quantum AI
One very large wildcard in artificial intelligence which could radically change every assumption that has been made about “pacing” in the sophistication of AI, job replacement, etc., is quantum computing. There are some scientists who don’t believe that quantum computing will ever actually take off - but companies and governments are watching it very, very closely.
And of course - it could be quite wonderful for science too - in terms of helping solve questions and making advances that may be too complex to fully figure out at present. Cancer, Space Exploration, Climate Change, etc.
Ex: Climate Change and AI - 2.17.19
https://www.oxfordstudent.com/2019/02/17/climate-change-the-case-for-artificial-intelligence/
Inspiration: If you like fiction, try reading the Quantum Spy: https://www.amazon.com/Quantum-Spy-Thriller-David-Ignatius/dp/0393254151)
By pacing, I mean that if/when quantum computing becomes more practical (billions are being poured into exactly this pursuit), the pace of increasing sophistication in AI and robotics may increase beyond anyone’s expectations. The existence of quantum computing is one of the reasons I think any prediction of something that “AI can never do” is very uncertain.
So my recommendation is try to keep an eye on quantum computing. Maybe by regularly looking at a site like Quanta.com. Or simply by keeping your eyes on sources of information like MIT Technology Review.
The world’s first quantum software superstore—or so it hopes—is here
Zapata Computing plans to build the algorithms for companies that want to experiment with quantum computers.
The Argument Against Quantum Computers
https://www.quantamagazine.org/gil-kalais-argument-against-quantum-computers-20180207/
The mathematician Gil Kalai believes that quantum computers can't possibly work, even in principle.
With all the changes that have already happened with AI, and the disruption that is expected to continue happening, it’s also important to note that there is a viewpoint that quantum computing in and of itself will be more disruptive than AI.
Quantum Computing, not AI, will Define Our Future
https://techcrunch.com/2018/11/17/quantum-computing-not-ai-will-define-our-future/
Studies
Studies that got my attention that I think are worth considering.
(World Economic Forum)
Machines will do more work than humans by 2025, says the WEF - Sept 2018
In less than a decade, most workplace tasks will be done by machines rather than humans, according to the World Economic Forum’s latest AI job forecast.
Notes from the AI frontier: Modeling the impact of AI on the world economy - Sept 2018
Artificial intelligence has large potential to contribute to global economic activity. But widening gaps among countries, companies, and workers will need to be managed to maximize the benefits. The role of artificial intelligence (AI) tools and techniques in business and the global economy is a hot topic. This is not surprising given that AI might usher in radical—arguably unprecedented—changes in the way people live and work. The AI revolution is not in its infancy, but most of its economic impact is yet to come.
Education - Implications of AI
Note: I am making Surfing the Tsunami available free to any educator, staff member and all of their students - http://tsunami.ai/edu
I believe there is an important need for schools and colleges to explore how AI and data science can help prepare students for jobs in the future. I believe it is an interdisciplinary question, and leading schools are blazing a path. The president of MIT called on the country and education to help students in all disciplines to become AI bilingual - to develop fluency in data and artificial intelligence. It’s not just a question for a computer science department. For example, a business department can bring business applications of AI and underlying analytics to the table. (Ex: digital marketing automation.)
If you are an educator, staff member, student or concerned citizen, some of these articles may be helpful in starting discussion around these issues.
Impact of AI on Higher Education
http://tsunami.ai/white-papers
> This is a pretty easy to read visual presentation I gave at a 2018 academic conference, based on the themes in Surfing the Tsunami. It might be helpful as a class presentation or just to get acquainted with the topic.
Rafael Reif, President of MIT
Fact: MIT is seeking to help students in every discipline to increase their technical skills and to leverage AI
Implication: you don’t have to be MIT to come to the same conclusion, and personally I think collaboration is necessary between schools, all their departments, companies and governments around the world to help students in any school to learn more, as well as anyone, of any age, who is not currently in school. That’s the main reason I wrote this book.
The President of MIT’s official release calling on education and government to support AI, as well as interdisciplinary AI education.
http://news.mit.edu/2019/president-reif-oped-federal-opportunity-and-threat-ai-0211
A related related opinion piece he wrote discussing AI bilinguality:
https://www.ft.com/content/24f18c28-2a39-11e9-9222-7024d72222bc
A nice video clip from Switzerland (brrr!) with the same message:
Joseph Aoun, President of Northeastern University
With Changing Students and Times, Colleges Are Going Back to School
https://mobile.nytimes.com/2018/04/05/education/learning/colleges-adapt-changing-students.html
(Comment from Todd: The kind of information pointed to by the excerpt below leads me to believe that facing this issue is an urgent question; the bolded statement (my bold) seems to additional confirmation that this is a now issue, not a future issue.)
—
That keeps Joseph E. Aoun, president of Northeastern University in Boston, up at night. While other presidents in local college towns worry about competing for endowments and enrollment, Mr. Aoun sees another threat: robots.
More than the latest polls, he is driven by a 2013 Oxford University Study that predicted that nearly half of the jobs in the United States are at risk of being taken over by computers within the next two decades. “We said if robots are going to replace human beings we need to help students to be robot-proof, and we built a strategic plan around that,” Mr. Aoun said.
That thinking positioned Mr. Aoun on the fringe of higher education strategizing just a few years ago, but he is now called on weekly to advise other institutions on how to help their students outsmart the workers of the future.
--
When it comes to lifelong learning, are universities the providers of last resort?
https://www.linkedin.com/pulse/when-comes-lifelong-learning-universities-providers-last-joseph-aoun/
> A good article that talks about the implications of AI on the job market as well as the need for lifelong learning for those who either never had college or graduated from college.
He even used the word tsunami!
--
It is no longer debatable that artificial intelligence will transform our economy; the only question is the precise size of the tsunami. Numerous studies predict that up to 50 percent of today’s jobs will disappear over the next two decades. In developing economies it could be up to 70 percent.
These same forces will create new jobs that we can’t even dream of today. But those jobs alone will not be sufficient: Just one in 10 workers are currently employed in fields that are projected to grow. Additionally, more than one in three existing jobs will require entirely different skill sets in the future.
--
Five Rules of the College and Career Game
https://cew.georgetown.edu/cew-reports/5rules/
As postsecondary education and training has become the most well-traveled pathway to middle class earnings, students, their families, and educators need to learn five rules of the college and career game. And sometimes those rules are contradictory.
> Report, a PDF, and interactive tool. Information on salaries of different majors, etc.
UC Berkeley’s Fastest-Growing Class Is Data Science 101
https://www.wsj.com/articles/at-berkeley-its-big-data-on-campus-1541066401
Government/Public Policy
Free for Government People: I am making Surfing the Tsunami available free to any legislator or government staff member, at any level, in any government in the world - and all their constituents: http://tsunami.ai/edu
Open Statement to Government: I believe it is crucial for any and every government to learn more about AI. If you know of any legislator who might be open to reading a book like Surfing the Tsunami, and how to get in touch with them, please let me know. For U.S. legislators I may be able to provide print copies, if not PDF copies, and I’m trying to find someone who could help me offer the book (and input) to every member of the Congress and the Senate. If you know of a relevant organization or are convinced of the importance of AI and you want to help, please let me know! http://tsunami.ai/contact
What to Review: in addition to this section, government staff and legislators will probably want to review the sections in the book about thought leaders, education, etc.
The President of MIT’s official release calling on education and government to support AI, as well as interdisciplinary AI education.
http://news.mit.edu/2019/president-reif-oped-federal-opportunity-and-threat-ai-0211
A related related opinion piece he wrote discussing AI bilinguality:
https://www.ft.com/content/24f18c28-2a39-11e9-9222-7024d72222bc
A nice video clip from Switzerland (brrr!) with the same message:
National/International
Kai Fu Lee’s New Book: AI Superpowers
I think Kai Fu Lee’s book is worth reading, to learn more about how AI works, the related issues, and opportunities for the future for individuals and countries. In the presentation I attended in Chicago in September of 2019, Kai explained that the publisher pushed for the stark national imagery on the cover. But as a person Kai Fu Lee is mild-mannered, and his perspective is very valuable. The book does touch on international themes - but the conclusions are not really “us vs them” - one of the most interesting conclusions is the interesting idea that governments can and should encourage the development of jobs requiring compassion and empathy. (See the CBS 60 Minutes video and articles mentioned in the Leaders section.)
https://www.amazon.com/AI-Superpowers-China-Silicon-Valley/dp/0358105587
Note: it’s the #1 release in Government Management, and #1 in other categories.
State Government
Millions of Californians’ jobs could be affected by automation — a scenario the next governor has to address
https://www.latimes.com/projects/la-pol-ca-next-california-work/
U.S. Government
Universal Basic Income
(ex: when job losses increase, could UBL be a solution)
Teaching AI
AI for Business (multiple sections including one on teaching AI)
High School - Advance Placement Computer Science (which is aimed at an interdisciplinary audience)
https://www.nytimes.com/2019/02/12/opinion/college-board-sat-ap.html
“A few years ago, the leaders of the College Board, the folks who administer the SAT college entrance exam, asked themselves a radical question: Of all the skills and knowledge that we test young people for that we know are correlated with success in college and in life, which is the most important? Their answer: the ability to master “two codes” — computer science and the U.S. Constitution.”
> Note: they are talking about all high school students.
Teaching AI - Practical vs Theoretical
Q: What would be the difference between using a pre-created library or set of functions, and needing to use more in depth math? Are you saying that for many high level, validated use cases, pre-created methods are sufficient, where only high level math is needed? Would it be fair to say that a deeper understanding of math is needed in order to customize for a particular context, or is the deeper understanding of math only needed for theoretical work where there is the need to develop new applied use cases?
A: In the textbook for my course, each ML algorithm is almost a line of code. The users only need to know what it does. Deeper understanding of math is of course helpful to understand the algorithms, but might not be needed. Understanding the math can enable more theoretical work.
Ex: http://www.jmlr.org/papers/v19/ and https://arxiv.org/list/stat.ML/recent
Coding
A small team of student AI coders beats Google’s machine-learning code
https://www.technologyreview.com/s/611858/small-team-of-ai-coders-beats-googles-code/
The success shows that advances in artificial intelligence aren’t the sole domain of elite programmers.
Fun
I’m on a quest to find games that can help people learn math, coding, and even neural networks. The pickings are slim but the possibilities are huge. Please use tsunami.ai/contact if you know of anything, or would be interested in helping to either test or develop something.
https://github.com/ypwhs/NNPlayground
Whitepapers/Presentations/PDFs
This section has some of the PDFs that are available at http://tsunami.ai/white-papers
Impact of AI on Higher Education
> This is a pretty easy to read simple presentation I gave at a 2018 academic conference, based on the themes in Surfing the Tsunami. It might be helpful as a class presentation or just to get acquainted with the topic.
Top 100 AI Companies 2018
> An interesting view of AI Startups and larger companies. Probably somewhat USA-centric.
Realizing the Benefits of Automated Machine Learning
> DataRobot is a company that is pursuing AutoML (not Google's AutoML, but their own variety) - in the hopes of making it easier to learn about and leverage AI at companies. DataRobot has a lot of good material and could help especially with people who don't necessarily have a computer science background.
Next Generation Digital Workforce
A 2018 report by Everest Group, targeting the enterprise, for WorkFusion, making the case for Robotic Process Automation.
Enterprise Artificial Intelligence (AI) Services 2018
> analysis by Accenture on enterprise AI
AWS (Amazon) Cloud Certification 2018
> I'm strongly in favor of certifications, I believe there should be multiple AI certifications (Please use the contact form if you're interested, if you know of one, or know anyone who'd like to help develop one)
> Amazon has a lot of educational material - at the present they have a "cloud" certification, which might be worth looking into. They also have AI training material, and a tool called Sagemaker. Amazon's products/training are one way to learn AI, like Google's similar tools/training material.
Market Guide for AI-Related Consulting and SI Services
for Intelligent Automation
> enterprise level guide to give a sense of the kind of AI companies and consulting out there.
Data Integration and Machine Learning: A Natural Synergy
> This is a nuts and bolts guide that gets deep into the weeds on how data integration is important for implementation of AI at companies.
ABSTRACT: There is now more data to analyze than ever before. As data volume and variety have increased, so have the ties between machine learning and data integration become stronger. For machine learning to be effective, one must utilize data from the greatest possible variety of sources; and this is why data integration plays a key role. At the same time machine learning is driving automation in data integration, resulting in overall reduction of integration costs and improved accuracy. This tutorial focuses on three aspects of the synergistic relationship between data integration and machine learning: (1) we survey how state-of-the-art data integration solutions rely on machine learning-based approaches for accurate results and effective human-in-the-loop pipelines, (2) we review how end-to-end machine learning applications rely on data integration to identify accurate, clean, and relevant data for their analytics exercises, and (3) we discuss open research challenges and opportunities that span across data integration and machine learning.
(McKinsey)
MGI Notes from the AI frontier: Modeling the impact of AI on the world economy - Sept 2018
Artificial intelligence has large potential to contribute to global economic activity. But widening gaps among countries, companies, and workers will need to be managed to maximize the benefits. The role of artificial intelligence (AI) tools and techniques in business and the global economy is a hot topic. This is not surprising given that AI might usher in radical—arguably unprecedented—changes in the way people live and work. The AI revolution is not in its infancy, but most of its economic impact is yet to come.
International AI - China, Finland etc
If you know of or find resources in other languages and/or for other countries, please let me know: http://tsunami.ai/contact
Finland’s grand AI experiment
Inside Finland’s plan to train its population in artificial intelligence
Inside the Chinese lab that plans to rewire the world with AI
Alibaba is investing huge sums in AI research and resources—and it is building tools to challenge Google and Amazon.
China’s leaders are softening their stance on AI
A year after announcing an aggressive plan to dominate artificial intelligence, China’s vice premier has called for international collaboration.
> I happen to believe that science has no borders, and that international collaboration is crucially important, especially if you are in the majority of people who believe that climate change is a big problem. (Wouldn’t it be nice to collaborate and help spare future generations from the consequences of climate change? And every other threat to health for that matter?)
Ex: Climate Change and AI - 2.17.19
https://www.oxfordstudent.com/2019/02/17/climate-change-the-case-for-artificial-intelligence/
Debate: AI Job Impact
https://www.wsj.com/articles/inside-the-new-industrial-revolution-1542040187?mod=mhp
—
(My Response)
To claim that “ it is less about replacing workers“ is misleading and inaccurate.
The biggest gap in this analysis is the lack of mentioning how many jobs will be lost from automation of both routine tasks and sophisticated analysis. Large employers in their own words point to this, such as the June 2018 report that Citi expects to shed 20,000 jobs.
https://www.ft.com/content/579c977c-6d73-11e8-92d3-6c13e5c92914
There is no mention of the many studies by PwC, McKinsey and others, who forecast millions of jobs to be lost, or the fact that there is consensus that there will be massive job loss but that estimates very wildly about how many and when.
Experts in the field like AI Investor and scientist Kai Fu Lee openly address the massive job loss and urge serious consideration at a public policy level.
New jobs will be created, but net job creation is not guaranteed; like a Great Recession in job loss but potentially far greater: it is very much about replacing workers.
Request: Feedback, Links, Resources
Q: Most importantly, what do you think?
Q: Anything I should add, change, improve?
Resources
www.techemergence.com - good source of news and information about the AI industry
VentureBeat AI Weekly Newsletter
https://venturebeat.com/newsletters/
MIT Technology Review
> publication, digital access, as well as some things you can subscribe to without necessarily paying.
www.quanta.com - quantum computing