Will artificial intelligence be humanity’s salvation or our undoing? An economist at MIT thinks it’s too soon to tell for certain — but he says we’re currently on the wrong track.
The problem isn’t AI itself, of course. It’s how we use it. According to professor Daron Acemoğlu, whose 2013 book Why Nations Fail was a Wall Street Journal bestseller, organizations that are good at applying statistical pattern recognition to large datasets gain too great of an advantage over consumers, workers, and competitors. Acemoğlu isn’t concerned with fairness for its own sake. He’s worried that AI, as it’s currently being developed, will lead to consequences that far outweigh the benefits of the technology in the long run.
The good news is that regulation can fix the problem, but only if we act soon.
The AI-Driven Marketplace
Online shopping seems like a well-functioning market where consumers have more than enough information to make a smart purchase. Acemoğlu acknowledges that AI can benefit consumers by helping companies make better products. For example, AI-based insights can be used to improve product quality and better meet customers’ needs. However, in a recent article in voxEU, Acemoğlu explains that he has found evidence of several ways the technology is detrimental to the market and heading down the wrong path.
The primary problem is that companies that collect a lot of data on their customers can also use AI-based insights to squeeze every cent out of each transaction.
Case in point, retailers use a technique called “price discrimination” to make as much as possible on each transaction by charging different customers different prices. The data gives retailers an indication of how badly specific customer segments need or want a product and how much they will be willing to pay, and they set their prices accordingly. Another technique involves showing personalized ads during “prime vulnerability moments” when the ads are likely to be most effective.
These techniques make business sense for the companies that use them, but they stifle competition and reduce the amount of money customers have to spend elsewhere. Most alarmingly, Acemoğlu notes that it also increases inequality and discrimination.
AI at work
AI is also changing the relationship between businesses and their workers. New inventions have been putting people out of work — and creating jobs for others — since long before the advent of AI. What’s different now is the scale and pace of the coming transformation.
AI is already changing how work is done in industries from healthcare to manufacturing to finance. Acemoğlu didn’t find evidence of a looming crisis of mass unemployment — at least, not anytime soon — but his research shows that AI-driven automation is poised to further increase economic inequality by forcing some workers into lower-paying jobs or out of work entirely. The result? Democracy itself could suffer if AI further “shifts the balance of power away from labor [and] towards the capital,” says Acemoğlu. In effect, AI could make entire classes of people dispensible.
It may be difficult to accept that one technology can have such an impact on massive, foundational institutions like democracy. However, we’ve already had a small taste of what can happen when AI-powered algorithms are unleashed on society, as a large experiment recently confirmed that Facebook’s algorithm limits users’ exposure to news told from a perspective they don’t agree with. This helped enable the spread of extremism and misinformation on social media.
Toward a Brighter Future
There’s no denying that artificial intelligence promises many benefits for humanity. In just the field of medicine, AI has already given clinicians tools to diagnose some conditions more accurately and treat them more effectively.
Then, how can we benefit from AI without suffering from too many of its harms? Acemoğlu says the answer is smart regulation. The worst-case scenario can likely be avoided if policymakers shape regulation to slow the use of AI for some applications, including automation and certain forms of data harvesting. Those rules have the potential to give ordinary people a chance to benefit from the AI revolution, not just the major corporations that currently drive its research and implementation.
Acemoğlu isn’t the only AI expert who’s trying to make algorithms work better for society. In an essay in the New York Times, former Facebook data scientist Roddy Lindsay says Congress can address the problem by using the law to change the platforms’ incentives. His solution? Make the companies liable for the content their algorithms push.