Sam Altman, on a global tour to charm national leaders and
powerbrokers, said in Paris that AI would not — as some have warned — wipe out
whole sectors of the workforce through automation.
"This idea that AI is going to progress to a point
where humans don't have any work to do or don't have any purpose has never
resonated with me," he said.
Asked about the media industry, where several outlets
already use AI to generate stories, Altman said ChatGPT should instead be like
giving a journalist 100 assistants to help them research and come up with
ideas.
ChatGPT burst into the spotlight late last year,
demonstrating an ability to generate essays, poems and conversations from the
briefest of prompts.
Microsoft later laid out billions of dollars to support
OpenAI and now uses the firm's technology in several of its products — sparking
a race with Google, which has made a slew of similar announcements.
Altman, a 38-year-old emerging star of Silicon Valley, has
received rapturous welcomes from leaders everywhere from Lagos to London.
Though earlier this week, he seemed to annoy the European
Union by hinting that his firm could leave the bloc if they regulate too
severely.
He insisted to a group of journalists on the sidelines of
the Paris event that the headlines were not fair and he had no intention of
leaving the bloc — rather, OpenAI was likely to open an office in Europe in the
future.
Exhausting
The success of ChatGPT — which has been used by politicians
to write speeches and proved itself capable of passing tough exams — has thrust
Altman into a global spotlight.
"Years from now, reflecting on this will feel very
special... but it is also quite exhausting and I hope life calms down," he
said.
OpenAI was formed in 2015 with investors including Altman
and billionaire Twitter owner Elon Musk, who left the firm in 2018 and has
repeatedly bashed it in recent months.
Musk, who has his own AI ambitions, said he came up with the
name OpenAI, invested $100 million in it, was betrayed when the company turned
itself from non-profit to profit-making in 2018, and has said Microsoft now
effectively runs the company.
"I disagree with almost all of that, but I will try to
avoid a food fight here," said Altman. "There's got to be more
important things than whatever he's going on about."
Instead, he wanted to focus on the mission of OpenAI, which
he said was to "maximise the benefits" to society of AI and
particularly Artificial General Intelligence (AGI) — the much-vaunted future
where machines will master all sorts of tasks, not just one.
He conceded that definitions of AGI were "fuzzy"
and there was no agreement, but said his definition was when machines could
make major scientific breakthroughs.
"For me, if you can go figure out the fundamental
theory of physics and answer it all, I'll call you AGI," he said.
A major criticism of his products is that the firm does not
publish the sources it uses to train its models.
As well as copyright issues, critics argue that users should
know who is responsible for answering their questions, and if those replies
used material from offensive or racist webpages.
But Altman argued the bottom line was that critics wanted to
know whether the models themselves were racist.
"How it does on a racial bias test is what matters
there," he said, deflecting the idea that he should publish the sources.
He said the latest model, GPT-4, was "surprisingly
non-biased".