Yet fiction will circle the globe halfway before the truth is revealed, and the story must permeate our collective unconscious fears about the threat of AI to humanity. The idea is gaining traction thanks to warnings from two of his “godfathers.” ” His two open letters on AI and existential risk.
Fears of rampaging gods and machines are being triggered, which are deeply ingrained in our culture, but we all need to calm down and take a closer look at what’s really going on here.
First, let’s acknowledge a group of computer scientists who have long believed that AI systems like ChatGPT need to be more carefully aligned with human values. They suggest that designing AI to follow principles like honesty and kindness makes it less likely that they will turn around and try to kill us all in the future. I have no problem with these scientists.
But in recent months, the idea of an extinction threat has become so well-established in public that if you bring it up at dinner with your in-laws, they’ll all agree on the importance of the issue.
At first glance, this is silly. This is great news for big AI companies, too, for two reasons:
1) It creates the specter of an all-powerful AI system that will eventually become more mysterious than we can expect to understand. That may sound intimidating, but the flood of purchases and deployments of AI systems is also making these systems more attractive. This technology may one day destroy the human race, but doesn’t it show just how big an impact it can have on businesses today?
This kind of paradoxical propaganda has worked in the past. Prestigious AI lab DeepMind, largely seen as OpenAI’s biggest competitor, is a research lab with the ambitious goal of building his AGI (Artificial General Intelligence) that could surpass human capabilities. started as When founders Demis Hassabis and Shane Legg first approached big venture capital investors like Peter Thiel for funding more than a decade ago, the technology’s existential threats didn’t hesitate about. In fact, they talked openly about the risks and got the funding they needed. (1)
By spotlighting AI’s ability to disrupt the world in an ambiguous way, we can let our imagination fill the void and speculate that future AI will have limitless capabilities and powers. It’s a great marketing strategy.
2) diverting attention from other efforts that could harm the business of the big AI companies; Some examples: The European Union voted this month on a law called the AI Act that would force OpenAI to disclose copyrighted material used to develop ChatGPT. (OpenAI’s Sam Altman initially said his company would “stop doing business” in the EU because of the law, but has since backtracked.) And advocacy groups recently told the U.S. Federal Trade Commission , has launched an investigation into OpenAI, asking it to urge the company to meet regulations. Authorities require AI systems to be ‘transparent and explainable’ [and] fair. ”
Transparency is at the heart of AI ethics, and an area where big tech companies invested more heavily in 2015-2020. At the time, Google, Twitter, and Microsoft Corp. all had strong teams of researchers investigating how AI systems like the one powering ChatGPT could unintentionally persist. I was. Prejudice against women and minorities, invasion of privacy, destruction of the environment.
However, the more the researchers investigated, the more it became clear that their business model was part of the problem. A 2021 paper by Google AI researchers Timnit Gebreu and Margaret Mitchell found that the large language models employers are building can be dangerously biased toward minorities. That opacity makes the problem worse, he said, making it more susceptible to abuse. Gebreu and Mitchell were subsequently fired. Microsoft and Twitter have also dismantled their AI ethics team.
According to Alexa Haggerty, an anthropologist and affiliated researcher at the University of Cambridge, this was a wake-up call to other AI ethics researchers. “‘You are hired to raise ethical concerns,’ she said, characterizing the tech company’s view of ‘but please stop raising issues we don’t like.’ I’m here.
As a result, there is now a crisis of funding and attention to the AI ethics field, confusion about where to go if researchers want to audit AI systems, and big tech companies looking to audit their AI systems. It’s made more difficult by the fact that it’s becoming more secretive about what the models look like. became popular.
It’s a problem even for people who worry about catastrophe. If these systems aren’t transparent and humans don’t have the expertise to scrutinize them, how will people be expected to control AI in the future?
The idea of solving an AI black box is often touted as near-impossible, but it may not be that difficult. In his May 2023 article in the Proceedings of the National Academy of Sciences (PNAS), the peer-reviewed journal of the National Academy of Sciences, the so-called explainability problem of AI is as unrealistic as many experts previously thought. showed that it is not. .
Technologists like OpenAI CEO Sam Altman who warn about the catastrophic risks of AI often do so in vague language. But if such organizations truly believe that their technology has the slightest chance of destroying civilization, why build it in the first place? It contradicts the long-term moral mathematics of Silicon Valley AI developers, who say that small risks should come first.
Not only would it be wiser to watch AI systems more closely now than to wring out a vague future apocalypse, but it would also put humans in a stronger position to prevent catastrophic events from happening in the first place. I can. But tech companies want us to worry about the distant future rather than want transparency about their algorithms.
When it comes to our future with AI, we must resist sci-fi distractions from the greater oversight we need today.
Details from Bloomberg Opinion:
• ChatGPT should go to college. Will OpenAI Pay?: Palmy Olson
• Breathless AI ‘extinction’ warning mistake: Tyler Cowen
• At the heart of AI is the world’s rule-maker over barrels: Dave Lee
(1) From Genius Makers: The Mavericks Who Brought AI to Google, Facebook and the World by Cade Metz, describing Hassabis and Legg’s early fundraising efforts.
This column does not necessarily reflect the opinions of the editorial board or Bloomberg LP and its owners.
Palmy Olson is a technology columnist for Bloomberg Opinion. She is a former Wall Street Journal and Forbes reporter and author of We Are Anonymous.
More articles like this can be found at bloomberg.com/opinion.
