B.I.S.S Research White Papers

AI For The Non Technical

Kevin Gray

Kevin Gray

I draw upon advanced methods from diverse fields such as Econometrics, Biostatistics, Psychometrics and Machine Learning extensively - for details see http://cannongray.com/methods.
As a marketing science person, I need a good understanding of what AI is and what it can do, especially in light of all the hype about it in recent years.

There’s a joke circulating in the data science community that if something is written in Python, it’s machine learning and if it’s written in PowerPoint, it’s AI.

I’ve now read quite a bit on this subject and am in frequent touch with AI specialists, and have briefly summarized the results of my informal research here.

There are many excellent books, articles, YouTube lectures and blogs on AI and topics related to it aimed at data scientists and AI researchers. Often, however, in the business world “AI” really means data mining and predictive analytics, and Data Mining Techniques (Linoff and Berry) is an excellent overview of that subject which will not go over the heads of most marketing researchers. It is nearly 900 pages long, however. Also excellent, though more mathematical, is Introduction to Algorithmic Marketing (Katsov).

There are many other books about AI written on a more technical level I have found helpful, for example: 

  • Artificial Intelligence (Russell and Norvig) – Considered by many data scientists as the “bible” on AI and more than 1,000 pages in length. Statisticians reading this book will note the many similarities between AI and statistics.
  • Reinforcement Learning: An Introduction (Sutton and Barto) -Regarded by many as the “bible”on an important type of machine learning heavily used in robotics and other applications of AI where learning by trial-and-error is critical. This is a highly technical book. Reinforcement learning is probably closer to what lay people think of when they hear “AI” than unsupervised or supervised learning. (See the Linoff and Berry book for more information on those two approaches.)
  • Deep Learning (Goodfellow et al.) – Widely-cited and essential reading for data scientists and statisticians who’d like to study these kinds of Artificial Neural Networks at a detailed mathematical level. The authors are leading authorities on Deep Learning.
  • Natural Language Processing for Social Media (Farzindar and Inkpen) – A thorough review of recent academic research on the use of NLP for text mining social media. Social media present special challenges for NLP (e.g., slang, poor grammar, sarcasm, impersonation). I have interviewed the first author here.
  • Machine Translation (Poibeau) – The best overview of automated translation I’ve read. Not extremely technical but might be a challenge for those will little background on this subject. Machine translations vary in quality and by language pair and direction of translation; for example, translating from German into English works better than translating from English into German.
  • Recommender Systems Handbook (Ricci et al.) – Recommender systems are now so ubiquitous that we scarcely notice them. This is a mammoth reference (over 1,000 pages) and a comprehensive mathematical look under the hood of recommender systems. I found the book both eye-opening and humbling.  

While I’ve learned a great deal from books such as the six listed above, they would be of little interest to people who are not marketing scientists and not working in data science-related areas. They all are quite technical. Confusing matters are the news articles, blogs, conference presentations and what I call airplane books which are superficial or even misleading.

So how can people not interested in the theoretical and mathematical details of AI learn about it? One source I can highly recommend – though admittedly I’m biased – is the MR Realities audio podcast series I co-host along with Dave McCaughan. Two discussions devoted to AI are: 

“AI: Reality, Science Fiction and the Future” (Mei Marker) 

“The AI Bubble” (Andrew Jeavons) 

Recently, I canvassed my LinkedIn connections for resources they would recommend to those who do not need to know the nitty-gritty of AI. That discussion can be found here. Listed below are the books, blogs and other resources they suggested. Their names are given in parentheses and I would like to thank them for their ideas and many thoughtful comments. Apologies to anyone who commented after this article went to press. 

Podcasts and Courses:

“Will AI lead the next revolution in healthcare?” (Mei Marker) 

“AI for Everyone” (Boris Ettinger) 

“Risto Siilasmaa on Machine Learning” (Thor Osborn) 

Brandon Rohrer’s series (Lars Øvlisen) 

Books and Blogs:

The Hundred-Page Machine Learning Book (Jan Voetmann) 

Data Science for Business (Robert Smith) 

Machine Learning: The New AI (Mahta Emrani) 

The Master Algorithm(Nayef Ahmad)

The Cartoon Introduction to Statistics (Andrew Silver) 

Algorithms to Live By: The Computer Science of Human Decisions(David Clarke) 

The simplest explanation of machine learning you’ll ever read (Lakshmy Priya) 

The Future Is Artificial (Ann Louise Sæmer) 

Numsense! Data Science for the Layman (Mark Bertens) 

I hope you’ve found this interesting and helpful! 

Kevin Gray

Kevin Gray

I draw upon advanced methods from diverse fields such as Econometrics, Biostatistics, Psychometrics and Machine Learning extensively - for details see http://cannongray.com/methods.