The chief executive of OpenAI, Sam Altman, faced questions about Artificial Intelligence (AI) at a hearing in the US Senate
The company created the app ChatGPT, which can write essays, scripts, poems, and solve computer coding in a human-like way
Altman told the US Senate his industry needs to be regulated by the government as AI becomes 'increasingly powerful'
Politicians were searching for answers on the potential threats AI poses and raised fears over the 2024 election
Several experts including Dr Geoffrey Hinton, the so-called godfather of AI, have recently raised concerns about the fast-developing technology
In addition to focusing on the threats, the witnesses are making sure to focus on the good that can be done for society with technological advancements in AI.
OpenAI founder Sam Altman hopes AI can "address some of humanities biggest challenges".
Senator Durbin is up next, and he says he sees a difference in the approach taken by AI compared to what happened with social media firms.
He says there seems to be a message of “stop me before I innovate again”.
He then turns to Section 230 – which allows social media companies to be held accountable for issues on their platforms – and asks how this might apply to AI.
Altman's response? He “doesn’t know” the answer to that yet, but that he would like to collaborate with the government to work it out. He adds that a “liability framework” around using AI technology is important.
Senator Hawley, a Republican from Missouri, asks whether Americans should be concerned that AI models can predict public opinion.
He suggests that organisations can use what they learn to "finetune strategies to elicit behaviours from voters", alleging that even Google Search influences undecided voters based on how it ranks searches and what articles come up.
OpenAI CEO Altman concedes that he is "nervous" about the issue.
His proposal? Companies should voluntarily adopt, or be regulated to adopt, "guidelines about what's expected in terms of disclosure from a company providing a model".
Image source, Getty ImagesAsked about what his biggest nightmare regarding AI developments would be, Altman says that "he expects there to be significant impact on jobs, but exactly what that impact will be is hard to predict".
He says it's important to think about GPT-4 "as a tool, not a creature".
"It's a tool that people have a great deal of control," on how they use it. He adds that it is a tool for tasks, rather than jobs.
Looking back through the history of technology, he says that as techology grows, our quality of life raises as well.
He adds that "he is very optimistic about how great the jobs in the future will be".
James Clayton
North America technology reporter
As the hearing began, Senator Blumenthal played a recording of what sounded like his voice. Except it wasn’t.
In fact, he used a combination of AI generated voice cloning and GPT-4.
He didn’t say the words, he didn’t write the words - but it sounded exactly like him.
Voice cloning has been around for years, however companies like ElevenLabs have popped up recently that use powerful AI and can be used by anyone. The tool can be trained on as little as a minute of someone’s voice recording.
As Blumenthal says - it sounds impressive. But someone could equally clone his voice to say he supported Vladimir Putin. The tech could fuel disinformation and undermine trust in politicians.
And there are also worries about how the proliferation of this AI could turbocharge fraud - with impersonation becoming more convincing than ever before.
Click the play button at the top of the page.
Connecticut Senator Richard Blumenthal used ChatGPT to write his statement
The hearing has now opened up to questioning.
The first comes from Senator Blumenthal who asks Altman if he thinks it’s a good idea to establish “scorecards”, similar to nutrition labels, around content that tell users how much the content they are using can be trusted.
Altman says he thinks “it’s a great idea”, and that independent audits are important.
He adds that the models “are getting more accurate over time,” adding that while mistakes are still made, he believes users “are more sophisticated” than we might think in spotting where mistakes may have occurred in AI technology.
We earlier heard from Senator Lindsey Graham, who refers to the rapid rate of technological advances.
Referring to sitcoms, he reminds people of how fast advancements have been made - he uses Seinfeld as an example.
Looking at the record of Congress, he says, "the Senate was not created for that purpose, but just the opposite: slow things down, take a harder look at it".
Graham says he is glad this hearing is taking place in a bipartisan approach.
"We're going to have to scramble to keep up with the pace of innovation," but "this is a great start".
There are a couple of terms the witnesses will keep coming back to during today’s hearing.
Let’s have a look at what each means:
Professor Emeritus Gary Marcus is the last witness to give his final statement before questioning begins.
He says we have built machines "like bulls in a china shop," and we cannot guarantee people are safe.
Prof Marcus says OpenAI's initial mission statement said they were committed to benefiting humanity, but now they are beholden to Microsoft - one of the company's big funders.
He adds that "we have unprecedented opportunities," but we're also facing "the perfect storm".
"We acted too slowly with social media," and the choices we make now "will have lasting effects".
Zoe Kleinman
Technology editor
Sam Altman is off to a confident start with his opening remarks. He’s reading from notes, and his tone is assured, but without a hint of Silicon Valley arrogance.
He, his company OpenAI and their product, ChatGPT, have already had several favourable mentions in the opening remarks – this is not (yet) a CEO under attack.
He is unsurprisingly optimistic about the benefits of AI but admits he also has concerns.
In his introduction, Senator Hawley described AI as “one of the most significant tech innovations in human history” but he asked whether it would turn out to be more like the printing press – beneficial, helpful – or the atomic bomb.
Altman responds – it can be a printing press, he says, but “we have to work together to make it that way”.
Government regulation is “critical” as AI tools grow more powerful, he adds.
In the past we have seen tech firms insisting they can regulate themselves.
Altman knows that nobody’s going to entertain that this time around.
Image source, US SenateMontgomery is the Chief Privacy and Trust Officer for IBM
The second witness, Christina Montgomery, begins her testimony by saying that AI is not new "but it's certainly having a moment".
She says recent breakthroughs in the technology and a "dramatic surge" in attention has rightfully raised serious questions about AI, notably its impacts on society, and what should be done about things such as bias, misinformation and misuse.
She says these are the "right questions" to be asking.
She adds that the role for the government to play a role has also not passed, and says government must work with the business community.
"Now is precisely the time to define and build the right guardrails to protect people and their interests," she says.
The hearing is starting with all parties seemingly on the same page - agreeing that AI must be regulated.
Altman explicitly states that government regulation is crucial as the AI sector develops.
ChatGPT's latest model was rigorously tested, he says, but "regulatory intervention by governments will be critical to mitigate the risks of increasingly powerful models".
He calls for licensing and testing requirements as one option.
Altman adds that "it's important that companies have their own responsibility no matter what the government does" and that AI "is developed with democratic principles in mind".
We're now hearing Sam Altman talk us through the positive advancements made possible by his company's tools. He is the CEO and founder of OpenAI, which owns ChatGPT.
He says his technology supports accessibility advancements and can help find solutions to "some of humanities biggest challenges, like climate change and curing cancer".
He says the current systems are not yet capable, but "it has been immensely gratifying" to watch many people around the world get value from his systems.
Altman: 'AI has potential to find cure for cancer'
James Clayton
North America technology reporter
Tech bosses can be wonderfully idiosyncratic - and Sam Altman is no different. He’s classically Silicon Valley to his bones.
He’s thoughtful, considered and calm under pressure - traits that should serve him well here.
But he’s publicly said he has faults as a CEO. He’s said before that he’s not a brilliant salesman or indeed advocate for the tech.
He’s also not naturally pugnacious - he’s more Jack Dorsey than Elon Musk.
I’ve watched many interviews with Sam Altman and he rarely gets flustered.
So it will be interesting to see how he handles such a long and intense grilling.
Image source, Reuters
Image source, NoneWe'll now hear from the witnesses.
Altman is the first to speak.
Image source, NoneRepublican Senator Josh Hawley is reading his opening statement next, and says that AI technology is "transforming our world right before our very eyes".
He says AI is "really like the invention of the internet in scale," adding that its impacts could be even more significant.
He compares it to the Atomic bomb, which he says was a significant technological advance, but that its impacts "continue to haunt us to this day".
He says that now is the moment to establish how this technology can be used for good of the American people.
He urges it to be used for the good of all people, not just the few.
Congress has a choice, Blumenthal says, as the rise of AI makes it ever more clear the potential risks of this technological development.
He outlines weaponised disinformation, voice cloning and deep fakes as some of those risks.
Comparing this to the rise of social media, he adds that it is important for the US government to take this seriously, and "not repeat our past mistakes" by waiting too long to regulate.
Blumenthal says that the public "has been fascinated" with ChatGPT and other AI tools, and that school homework and op-eds written by the technology feel like novelties.
But he adds that it's clear they are "no longer fantasies of science fiction".
He calls the advances in AI technology encouraging and hopeful, but warns they also bring potential harm.