The current boom in generative artificial intelligence driven by tools like OpenAI's ChatGPT, which can seemingly produce organic text, images and more, represents a leap forward both in what the technology can do and its potential to influence everything from the internet itself to the minds of the people who use it.
The technology is having a moment, with the potential to overhaul millions of jobs while some experts are urging that it is accelerating too rapidly, although it has been in development in some form since World War II and the early days of the Cold War.
But there are still far more questions than answers about what the rapidly evolving technology will mean for the lives of everyday people as it is already becoming ubiquitous in search tools and other applications. So we asked two experts in the field — Rob Reich, associate director of the Stanford Human-Centered Artificial Intelligence institute, and Matt White, an AI researcher and UC Berkeley faculty member — about how the use of AI could play out for society, what it means for generations that will grow up with it, and much more.
AI has been called "the most powerful foundational technology" ever to have existed. Accurate?
"I would restrict that to maybe 'the most powerful technology of the next decade,' as opposed to the century or some other extended period of time," White said. "Because, honestly, the Internet was pretty powerful. The web was pretty powerful. And it changed a lot around the way we work and the way we live. So I would see AI being on the same level as those technologies."
"I would not necessarily agree with (former Google Chairman) Eric Schmidt's characterisation that it's the most powerful or most important technology." Reich said. "But if you agree, or even if you're a few degrees shy of agreeing, if it's that significant, then democratising access to it should sound like something to be worried about rather than something to celebrate. ... AI tools and technologies, they're relatively cheap. They're often available in open access, open source modes, and (if you agree with Schmidt) it would be, to my mind, it's akin to saying at the beginning of the nuclear age, what would it be like if everyone could play with plutonium and uranium?"
How will generative AI change education and how we learn to read and write?
"I think the consequence of this is that if people rely on machines to do their writing for them, then we face a generation that never learns how to write clearly or think clearly in the first place," Reich said. "If you've already learned how to write, it might be a help, but for an eighth-grader, it's a cheating machine that robs you of the ability to learn how to think and write well."
An example is White himself, who is working on a book about generative AI and uses the software as an aid. "I use ChatGPT to ideate or even to summarise my text, like, 'Rewrite this for me in a different way,' " White said.
But the technology has the potential to change how future generations use the internet, White said. "I call this search to chat, which is like you move off of Google and just looking and sorting through answers and spending so much time trying to find the answers you're looking for," as opposed to using generative AI "to be able to ask a question and get a very deliberate answer."
There are limits and complexities to that change, though.
"Right now, these large language models don't have citations, which is a big problem," White said. "They should be able to tell you where they've got that information from because of the way they're trained and the way they learn. They can't do that. And so that's something that has to change, and that can be built into systems as these systems get perfected."
Generative AI models are also prone to "hallucinate" by presenting inaccurate information as fact, another significant issue that needs to be worked out before the search to chat changeover can take place, White said.
What are the best- and worst-case scenarios for generative AI's integration in everyday life?
"The most optimistic scenarios are that AI is like the equivalent of the Industrial Revolution," Reich said, creating "exponential increases in productivity in the economy. It accelerates scientific breakthroughs, medical breakthroughs. It frees up human beings from having to do drudgery work, liberates people from unpleasant tasks and massively increases the size of the economy."
White mostly agreed: "The best scenario is one where we can enjoy life more and then not have to worry as much about work, create a more equal, balanced working life," while allowing the technology to "help us learn better, help us enrich our experiences, that sort of thing."
While the technology's role is unlikely to result in either of the most extreme situations both experts outlined, the worst-case scenario is unsettling.
"A pessimistic scenario is, inversely, this is an existential threat to humanity and that runaway AI that hasn't been harnessed to support human interests will turn out to subordinate humans to the machines' own interests, the artificial machine intelligence. And it's nothing short of an existential threat to the existence of humans," Reich said.
"Say you have an unconstrained machine that decides... to make decisions about trading in the marketplace. It could create unbelievably adverse conditions for the market where no human being, as it were, could reel it back in because it was immunised from any human control if it was a genuinely autonomous decision-making machine, an autonomous system."
White said the most significant misuse of the technology could come from the possibility that humans will not be able to discern what is a genuine image or video and what is AI generated, what he refers to as "synthetics." One example is a photo-realistic picture of Pope Francis that recently circulated online and turned out to be created by an AI program.
For example, that technology "can be weaponised by Russia to create synthetic media that pushes their agenda," White said, adding, "if nation-states have the tools to be able to create full synthetic video, now what happens is there's no truth in digital media."
How will the technology affect jobs and the job market?
One example is the software-programming industry itself.
"We have to ask ourselves in software development, 'Does it even make sense that we're writing all of this code?' " White said. "Why can't we just interact with machines and natural language?"
He envisioned a future in which a person can dictate to an AI model what they want their website to look like, "and then you just sort of sit there and customise things."
"At least in the short term, it's very complex to be able to go in and create an entire... website and all of the nice features of the website and all of these things just by generating code through text," White said.
But more remedial coding work could be supplanted by the technology.
"That means elite programmers will be in high demand. But I don't have to go and offshore a bunch of this development, because I can just do it here and automate a lot of the remedial type of development," White said.
On a larger scale, that kind of job displacement could have unintended consequences, according to Reich.
"I think the very idea that companies that create AI technologies and that sell them in ways that are meant to automate labour or replace labour are not taking account of a wider dynamic that takes place when human beings are fundamentally insecure about their economic stability or well-being," he said.
"That's a surefire recipe for political instability and populist revolts on the left and the right (and threats) to the stability of democratic institutions," Reich said. "So it's not merely a question of displacing labour. It's a shock to the political system how rapidly these technologies are changing the very nature of employment. And the same is true for education." – San Francisco Chronicle/Tribune News Service