AI: Game Changer, or A Solution Looking For A Problem?

Week after week, you come across news about disruptions in the automobile industry. It manifests in many forms ranging from new entrants in self-drive cars to environment-friendly electric vehicles, to ride-hailing operators turning to electric vehicles, or the highest decibel - autonomous vehicles. 

While research in the space of Autonomous vehicles is unraveling significant progress, and different companies are making competing claims on incremental success - there is a growing concern on the judgmental capabilities of these Artificial Intelligence machines. The question is - if the car finds itself facing choosing between two or more potentially fatal options - who would it choose? For instance, if the autonomous vehicle notices a young healthy man and elderly person, on its track who would it hit? How would it behave when choosing between a human and an animal. This is illustrated well with plenty of variations of the dilemma in The Moral Machine Experiment* an MIT study based on the Trolley Problem. 

Before getting to the results, you can play the Moral Experiment yourself.

Source: Nature

What this image tells us is that you are likely to be relatively safe from being on the wrong end of a moral decision conducted by a machine if you do not belong to one of the following demographic groups:

-Male

-A passenger

-Unhealthy

-Poor

-Unlawful

-Elderly

-An animal

There are certain variants based on which ethnic group you belong to. For simplicity, this article has left that out.

In short, it’s a question of ethics.

With the ethical questions looming large, and the fatal accidents where each of Uber and Tesla have been involved during tests - there is a need for rigorous testing. That begs the question - when would fully autonomous cars would be available? John Krafcik, Chief Executive of Google Waymo, believes, “It'll be decades before autonomous cars are widespread on the roads -- and even then, they won't be able to drive themselves in certain conditions.”

“It'll be decades before autonomous cars are widespread on the roads -- and even then, they won't be able to drive themselves in certain conditions” - John Krafcik - Chief Executive, Waymo

Given the technology constraints and viability challenges, “Every self-driving car on the road today is worse than a human,” says George Hotz, CEO of Comma.ai. “... So with a human, we believe these systems are safer than a human alone. And they certainly can be more convenient.” In a ride-hailing business, where autonomous cars are likely to make a significant impact, human drivers aren’t going away anytime soon. The technologies would only load the costs further, hence making them financially unviable.

Hypothetically, even if 100% autonomous cars (that is, without any human monitoring) were to make it to roads, where there is an alternative of human drivers, the human driver’s performance would be compared to the robot. This is a classical paradox where the human may be compelled to compete with the robot. 

A Cornell-led research team had humans compete with robots doing rote tasks to win cash prizes – and found that when the robot was doing well, the people tended to expend less effort, see themselves as less competent, and have a distaste towards the robots. At scale, this could have harsh social ramifications.

Artificial Intelligence has existed for decades, since 1956 when John McCarthy coined the term at MIT. Ever since the term has evolved but it has assumed wide acceptance and captured the imagination of scientists and innovators across disciplines and industry domains.

There are numerous use cases where AI is being deployed to solve unmet requirements, and this article focused on Autonomous Mobility - an extremely hyped application.

While there is a buoyant excitement among large sections of the industry and investors about the promises of AI, there are plenty of examples of the failure of AI to unleash the benefits it was meant to. For example, a World Economic Forum session “AI has huge potential – but it won’t solve all our problems” attempts to tone down the expectations of AI. An example illustrates how the healthcare industry has realized that AI cannot be the panacea for all problems. It cites the IBM Watson for Oncology programme - a specialized module of AI designed to help doctors treat cancer. In spite of the product’s proposed capability to deliver the best recommendations, it failed to gain the trust of expert oncologists. 

A lot has been written or said about the capabilities of AI. In a dog-eat-dog world, where investors are bullish, hundreds of AI start-ups trying to solve different business problems, are vying for their share of the pie. Over 3600 AI startups have raised $ 66B in funding since 2013, of which $ 7.4B has been raised in Q2’19 alone by 488 startups.

Source: CB Insights

The data speaks volumes about the excitement that AI has generated around the globe and the promise it holds out, in just the last 6 years. It also indicates the general acceptance of the technology among entrepreneurs, investors and industry alike.

A joint MIT Sloan Management Review and Boston Consulting Group survey of 2500 executives - Winning With AI - cites increasing awareness among leaders of and concern with competitors’ use of AI. However, The State of AI 2019 Divergence, a study of AI startups in Europe by MMS revealed only 60% of the startups – 1,580 companies – there was evidence of AI being core to a company’s value proposition. This raises a question if the opportunity is real or hype.

Thousands of executives call AI a game-changer. On the other hand, futurists call Advanced AI an existential threat for humanity. Tesla CEO, Elon Musk has been warning about the ethical implications of AI.

“AI is a fundamental risk to the existence of human civilization.” - Elon Musk

What’s your take on AI? 

Game changer? Or, A solution looking for a problem?

Further Reads: 

• *The Moral Machine experiment (Original Paper, Nature) - https://go.nature.com/2JVbPFz
• Experts Predict When Artificial Intelligence Will Exceed Human Performance - https://www.technologyreview.com/s/607970/experts-predict-when-artificial-intelligence-will-exceed-human-performance/
• Google DeepMind - https://deepmind.com/
• Elon Musk: Humanity Is a Kind of 'Biological Boot Loader' for AI https://www.wired.com/story/elon-musk-humanity-biological-boot-loader-ai/

Disclaimer: Views expressed in the article are those of the author, and do not have any bias or prejudice to any organization or professional group he may be part of.