Material is laid out in five simple parts titled, Prediction, Decision Making, Tools, Strategy, and Society, that were all very easy to follow. The authors appear to be moonlighting as business consultants, and so I can understand how having such a book published will help them with their street cred. They seem to be quite pleased with themselves in figuring out that AI is mostly prediction (rather than pure intelligence). They keep reiterating this point umpteen times, though it is not necessarily their own original discovery. Some ideas resulting from this understanding are indeed good & interesting. For example, the cost of computing going down and the quality of prediction going up considerably, can potentially transform business models. In case of Amazon, they may be able to predict exactly who will need what and when, in addition to their affordability. If this prediction is close to 100%, Amazon can start shipping items to people on their own and setup the logistics needed to collect very rare returns if/when needed. Thus, their business model may flip from shop first & ship next to ship first and collect returns. While this is a nice change in the business model that may make business leaders salivate, describing this insight takes up the first ~40% of the book, that covers Prediction & Decision Making parts. Perhaps the authors know that some/many of their intended readers won't even read this book in full or be so distracted that they won't follow even these simple discussions and so they summarize the key points at the end of each chapter. One key point (page 68) reads, "Humans make poor predictions, overweight salient information and do not account for statistical properties. Many scientific studies document these shortcomings across a wide variety of professions. The phenomenon was illustrated in the feature film Money ball." While that is all correct, is this the level of summary that should be provided? It is either meant for "Future business leaders still in high school" or else, we need to worry about the quality of our business leaders' natural intelligence!
Throughout the book authors touch upon every conceivable area that is connected to AI. This includes Amazon's Echo, robots used in their warehouses, Australian mining industry, self-driving cars, Google/Microsoft type companies' language translation tools, iPhone, Siri, and so on. There are quotes by Nobel laureates, jokes, 2x2 business charts, couple of graphs. But everything discussed is quite superficial, with none of the discussions spanning even couple of pages in the book. The advice they dispense also seems too simple. For example, one of the key points listed is that "C-suite leadership must not fully delegate AI strategy to their IT department". Chapter 16 opens with what is presented as a brilliant insight provided by one of the authors. Apparently, an early-stage ML company was trying to deliver a disease diagnosis tool to doctors and were struggling to get all the approvals needed to be able to do that, since they are basically doing what doctors are certified to do. The author, who is a business consultant had suggested that they deliver a probability number indicating how likely the patient has the disease and leave the yes/no diagnosis decision to the doctor. While we can agree that it is the right thing to do, is this a brilliant insight worth hyping as the opening paragraph for a chapter titled "When AI Transforms Your Business"? Reminded me of the recent John Oliver's Last Week Tonight episode, where he discussed business consultants. Watch 7:30 to 8:02 of this episode. As John Oliver himself says in that episode, I do have friends who are business consultants that are brilliant and who certainly add value to the business they work with, though that may not be the case always!
The last part titled "Society" is limited to just one chapter of about 10 pages (which is pretty much the topic of my entire 1.5 hour long lecture on Ethics & Emerging Technology). Of course, their answers and conclusions are aligned with my beliefs, which is heartening. For example, AI/ML has a lot of potential for increasing income inequality and so societies need to regulate/manage it well. Better AI performance may mean less privacy, without proper regulation, some companies may end up controlling everything in our lives, from production to distribution of all the goods & services. Hopefully this consistent view evolving on multiple fronts will help their intended readers get educated and aligned on these notions.
I was reminded of watching two travel documentaries way back in the early 90's when I was planning a visit to the UK. Since touring Wales was on the cards, picked up two VHS tapes about Wales tourism from the local public library. One talked about each attraction for 5 mins, provided a summary and a slide, listing bullets of the places/things one should see/do. Then moved on to the next part of Wales. Perhaps it was quite useful/productive to potential tourists taking notes. But the other video tape presented Wales beautifully like poetry. There were no bullet lists or take away summaries. I enjoyed that version so much that I watched it couple of times before returning the tape to the public library. This book belongs to the first kind. 
No comments:
Post a Comment