ChatGPT predicted Rishi Sunak’s speech about AI – here’s how it did | Science & Tech News

Rishi Sunak has said AI could be as transformative to British society as the Industrial Revolution.

The prime minister talked up the potential of the technology in a speech to mark a government report into the opportunities and threats it presents.

Ahead of the address, Sky News asked ChatGPT to predict what he was likely to say.

Politics latest: PM says AI can be ‘co-pilot’ for jobs

OpenAI’s chatbot is trained up to January 2022, before Mr Sunak entered Number 10, but we told it he was prime minister and asked – based on what it knew about him – what he would say about the risks of AI and how the world should approach regulation.

Economic implications

ChatGPT said as ex-chancellor, Mr Sunak may “highlight the economic benefits and challenges posed by AI”.

“This would include discussions on job creation and displacement, the need for reskilling, and how AI can drive efficiency and innovation,” it added.

The prediction proved to be a pretty fair reflection of the PM’s stance.

He said AI would have a similarly transformative impact as the Industrial Revolution, but was keen to stress how it would create new jobs rather than just cost people their existing ones.

“Everybody in our country can benefit,” he insisted.

Balanced approach to regulation

Given his “Conservative Party background”, ChatGPT said the PM may “emphasise the importance of a balanced regulatory approach that ensures innovation is not stifled.

The chatbot nailed this one – the PM said despite the concerns, the UK has always been a country that encourages innovation and will be no different with AI.

He said a UK taskforce and AI Safety Institute would help ensure models undergo proper safety assessments and ensure the government stays ahead of any risks.

But he also said he “won’t hide” the risks; from misinformation and bias, to far more profound dangers like humanity losing control of the technology.

Ethical implications

ChatGPT thought the speech would mention concerns like bias, privacy, and decision-making.

It also said the Tory leader “might argue for the development of guidelines or standards that ensure AI operates transparently and ethically”.

He did cite bias as one concern surrounding AI, though there was nothing too specific on guidelines.

But in answer to a question from Sky News about what restrictions could be put on AI developers, he suggested he would speak to tech leaders about it at next week’s Bletchley Park safety summit.

It would not be right for companies to “mark their own homework”, Mr Sunak added.

Collaboration

“Given the international nature of AI development, Sunak would likely call for global collaboration,” said ChatGPT.

This could include countries establishing “common standards and best practices” that aids all of humanity.

ChatGPT was right on the money here, with Mr Sunak saying there must be an “international approach” to safety.

He said the work of the AI Safety Institute will be “available to the world” and confirmed China was invited to the UK’s AI Safety Summit next week, despite opposition within his own party.

Education and skills

ChatGPT predicted an emphasis on education and training, ensuring the British workforce “is equipped with the necessary skills to work alongside AI and harness its benefits”.

It wasn’t until questions from the media that Mr Sunak really got into this topic.

He said all new tech brings changes to the jobs market, and the best way to prepare the country is via education, whether that’s at school, college, or through an apprenticeship.

Safety

This may have been towards the bottom of ChatGPT’s predicted topics list, but it was the focus of the speech.

The chatbot thought the PM would discuss “the potential dangers of uncontrolled or malicious AI, emphasising the need for rigorous testing, oversight, and measures to prevent misuse”.

Mr Sunak said he wanted to address safety fears “head on”, including the possibility that AI could make it “easier to build chemical weapons” and allow terrorists to “spread fear on an even greater scale”.

Humanity may even lose control of the technology altogether, he added.

Public engagement

The chatbot suggested Mr Sunak might “advocate for increased public engagement on the topic of AI”, such as through consultations.

There was no mention of any such events, with his message to the public much more “you can trust us to protect you”.

Mr Sunak was more interested in talking about co-operating with other world leaders and AI developers, starting at Bletchley Park next week.