I am returning from a break in my blog writing to address a book that profiles a game changing technology wave that will be more profound than any that came before it – including the PC, mobile or the internet.
The technology is AI algorithms, or “algos”. They build on all preceding platforms but are the only type of technology developed that have their own learning curve. Algorithms are rapidly replacing human labor in most areas and are accomplishing in many fields what humans can’t – learning, reprogramming themselves, and teaching us about patterns they are recognizing in the universe to solve specific problems.
The catalyst for this explosion of successful algos is the ever increasing amount of data becoming available to process. FMRI scans, sensor devices, DNA testing as well as mobile internet and social data are all generating a tremendous amount of digital information that can feed algos. This data helps them find patterns in the real world we could never see before.
After a wrong turn in the 1980s, AI technology now returns with an ever increasing number of success stories – including Amazon book suggestions, Google text search, self-driving cars, Netflix movie suggestions, etc. As futuristic as it seems, biotech robotic scientists today are able to create hypotheses, design tests, carry them out, and analyze the results at a pace never before possible. The output creates even more data for future algos!
The Master Algorithm explains why the pace of algo use is accelerating (and will continue to do so) while also demystifying the technology behind this transformation. The book describes the five major tribes of AI algos and shows examples of how they have been used with impressive results. The author goes far beyond that however – forecasting that these tribes will converge and ultimately form a Master Algorithm that will (with enough data) be able to derive all knowledge, past, present and future. He believes this algo exists, but remains undiscovered.
In the spirit of Ray Kurzweil’s, “The Singularity is Near”, Pedro Domingo proposes a future vision that is so transformed from today by algos that it is at first hard to believe, but convincing by the time you are finished. The book joins a growing chorus of futurists that see a future where humans will be challenged to compete with algos, and suggests we should instead relax and let them do the work. In this future, work may no longer be the primary way to secure resources for food, shelter, and a living for the majority of the population. Time to start finger painting! To show his commitment to this cause, Pedro finishes by providing open source access to his algorithm, – as close as he has been able to get to the Master Algorithm – for others to achieve the final model itself.
Why Listen to the Author?
Pedro Domingos has been tracking the evolution of this market since the 80s. He received his M.S. in Electrical Engineering and Computer Science (1992) from IST, in Lisbon and an M.S. (1994) and Ph.D. (1997) in Information and Computer Science from the University of California at Irvine and joined the faculty of the University of Washington in 1999. His is a prolific writer on AI and has online courses and several TedTalks.
He has seen the rise of the five AI tribes over time, diagnoses their strengths, where they work best, and stumble. He does a masterful job of explaining the benefits of each and how they can be stitched together to address problems in ways not possible before.
Next week
We will review the five AI tribes and discuss some use cases where they are applicable.
As a heads up, here is how he classifies them:
Symbolists, Connectionists (Neural),Evolution (Natural selection/ Genetic programming), Naïve Bayes and Analogizers
Comments