Here’s The Real Reason You’re Terrified Of The $1.2T AI Industry But Don’t Yet Truly Know Why

Last night a very intimate and provocative conversation around Artificial Intelligence took place in Paris, France at the American Library in the distinguished seventh arrondisement. The timing was quite fitting given a glowing, low, full moon under which the city which is also currently hosting an exhibit at the Grand Palais entitled Artists and Robots which is an impressive demonstration of art created by more than 40 artists with the assistance of robots enhanced with Artificial Intelligence.  Certainly, worry abounds regarding such collaborations and more as the expanse of AI barrels ahead. Thus, this particular talk on the AI’s impending impact on the future of business and humanity offered truly breakthrough thought-leadership pertinent to every business leader today.

Entitled “AI And The Future Of Humanity,” the event was produced by Ivy Plus European Leaders in partnership with UC Berkeley, UC Davis and featured two extremely prominent speakers. The first was internationally renowned AI pioneer Stuart Russell, Professor of Computer Science at UC Berkeley, Founder of the Center for Human-Compatible Artificial Intelligence and former adjunct professor of neurological surgery at UCSF and co-author of the textbook Artificial Intelligence: A Modern Approach used in more than 1350 universities in 119 countries.

The second was  Gerard Berry, Professor at Collège de France, Chaire Algorithmes, Machines et Languages andCNRS Gold Medalist.  He is also Director Research INRIA, the founder of Esterel Technologies and inventor of computer languages and programs with vast applications, such as enabling the Airbus 380  He is the author of, among other books of l’Hyperpuissance de l’informatique: algorithmes, données, machines, réseaux.

Each of the two men gave a presentation regarding their complimentary takes on A.I. While Berry’s was quite educational from a historical and basic education perspective, Russell’s was a potential game-changer for any and every business leader today.

He began his discussion by citing the fact that via AI, a computer taught itself how to play both chess and shogi to beat both the best human players and all previous programs in 24 hours flat and that it is such tech capability that has lead to the major frenzy that we see today with countries investing billions into efforts surrounding AI.

Yet Professor Russell’s astutely pointed out that there is a large question within the AI community about how such monetary allocations will be spent given the small number of people currently adept at anything even remotely resembling AI. Thus began the evening’s rabbit hole descent into myth versus fact as it pertains to everything from the speed at with AI will fully penetrate society to the capability with which machines truly have the capacity to out-perform human beings and take over the world.

This was about dissecting the hype.

In essence, business leadership today should understand the following takeaways, as they pertain to AI at the moment.  First, machines are ideal for usage in terms of collecting and parsing data.  The application of AI, for example, in going through massive amounts of satellite data that would typically take thousands of man-hours to complete or mining of books and various other media is ideal work for machines.  But when it comes to other elements, there are still breakthroughs which need to occur. Right now we are at the handing-over of monotonous work stage (blue collar and white collar jobs alike, mind you.)

Professor Russell stressed that the actual ability to plan and strategize is still very much under human domain. He noted that during just that evening alone that each of us in the room would perform more five million motor actions toward planning and execution. Score one for Team Human. The current ability to plan over the long-term is still missing from AI and that which human intelligence exceeds. We are also better at inventing given that, according to Russell, such activity is kicked off by cultural interaction (though one could argue that the machine would, in time, be interfacing with a collection of culture since the advent of mankind).

However, he equally cautioned that a more benign or dismissive view of AI is equally out of balance, citing the fact that after Lord Ernest Rutherford and Einstein declared at one point that it would be impossible to extract energy from atoms to which a noted scientist promptly left the talked and went home and invented neutron-induced electron chain-reaction the very next day.

Russell says that more balanced view is to try and truly see AI as a step-change in civilization since any significant amount of intelligence is what expands all civilizations.  In fact, he proposes that AI will increase standards of living across the globe by trillions of dollars, but that the downside could be the end of the world if humans lose control to the machines which they have created. Yes, you just read right. A veritable chess match, if you will, between humans and machines for which the entire world could be at stake could take place.

Thus, he is proposing a radical, new standard in AI. First, programming for a truly specific purpose needs to be mandatory in each case where AI is used. Right now, they are rather generally broad-sweeping. Second, the machine must be programmed in such a way that it can be turned off. Currently, machines know that in order to achieve the command set that is must, at the very least be powered. If that option cannot be overruled, well, think I, Robot in a very real way.

Russell says that AI is also currently being defined based on the idea of objectives and solutions to optimize them. “This is a fundamental mistake and could put our very civilization at risk,” he noted.  “So I am now advocating that there should always be two considerations first. One, the outcomes should always have the stipulation of ‘satisfaction of human preference’ attached each and every time.  Two, the machine does not, should not know what those preferences are. This prevents any type of what we would consider in humans to be psychopathic behavior carried out by the machine.”

Indeed, the machine would then always defer and also allow itself to be switched off, keeping humanity in tact.

And there is an incredibly strong economic incentive to get this right. Russell gave the example that if there are too many further additional reports regarding self-driving cars crashing or running into people, the entire industry will be defeated. “Right now, not a single car is even programmed with the preference, or understands that people like to be alive, first! It’s madness to work this way,” Russell exclaimed.

However, what is important to consider around this notion of preference can not only differ between cultures but within cultures.  While Russell wants to ensure that that AI is always set to defer to human preference so that our very civilizations exists, this is, arguably, a 30,000-foot view from which those already in this culture who are comfortable, leading are seeking to maintain current standards which they want to preserve. Where things could get sticky, and perhaps more pressing, is if “preferences,” such as African-American males are generally dangerous and should be all be incarcerated somehow gets looped into the AI system based on what some human “preference” is already seemingly acting out and for which policymakers such as Senator Cory Booker  are highly concerned, let alone the end of the world as we know it, though important.

Also discussed that evening was the fact that even if good intentions are developed, they may also not be sufficient.  Now more than ever will it be paramount to have diverse teams in technology.  For example, Berry cited the fact that only those with medical expertise were recently drafted to create new software for a certain pacemaker. It was later discovered all sorts of tech “holes” were in the program since no security tech experts were part of the original team. Thus, if programmed in a diabolical manner through these “holes,”  it was found that the pacemaker could act out the termination of the actual user rather than preserving his or her life. Yes, you just read right, again.

And this is only the beginning of the hotbed of challenge and foresight that business leaders will need in the new era of AI. Great standards and consensus will be needed, period.  This is an opportunity to learn how to better manage our lives, minds and business to create balance rather than rampant mayhem or complete suffocation of a powerful level of technology, period.

Berry cautioned, “There is a lot of economic rush toward AI at the moment.  If you remember the gold rush of the west, not that many people actually found gold. We all just have to approach this opportunity with a very, very thoughtful business strategy so that we can reap the benefits.”  (Forbes)