A former code breaking headquarters in Britain is hosting senior global figures next week as they try to protect tomorrow’s world from the dangers of artificial intelligence – but AI is already shaping our lives today.
Drawing on findings by spy agencies, official UK analysis says we are already “in the midst of a technological revolution” thanks to AI, which can, for example, produce increasingly high-quality content and score highly on school exams.
Prime Minister Rishi Sunak, who is leading the push for global protections on AI, believes it eventually will bring a “transformation as far-reaching as the Industrial Revolution, the coming of electricity, or the birth of the internet”.
His aim at the Bletchley Park summit is to get the world singing from the same hymn sheet about the risks of AI, which he says could include terrorism and weapons proliferation if the dangers are not held in check.
While the most outlandish AI predictions remain a matter of science fiction, software such as ChatGPT has made it increasingly visible and fuelled fresh debate about how to handle the technology.
The National has been hearing from start-ups, experts and political sources about how AI is already transforming education, medicine, law enforcement and countless other sectors.
So, what has AI ever done for us?
If you use Google Maps when you drive, your phone may be gathering data to help train the app to predict journey times.
Brake hard, even drop your phone, and AI can take it into account to forecast future trips – although this can clearly lead to errors.
Google Maps has also been trained to recognise things like pedestrian crossings and speed limit signs. An AI system is used to steer driverless buses that launched across Scotland’s Firth of Forth this year.
Dubai Taxi Corporation revealed this week that it has started using AI to track the performance of 14,500 drivers and 7,200 vehicles.
Scientists want to create a virtual copy of the human body – your “digital twin” – so doctors can personalise your treatment. That is still in the future, though.
What is already happening is that drug companies are using AI to discover new types of medicine.
One company called Pangea Bio combines modern software with traditional medicine. It asks AI to crunch through information on plants with therapeutic potential, and make data-driven decisions not available to healers long ago.
“We can generate data at a pace and quantity and volume that was not possible before,” the company’s head of AI, Sona Chandra, told The National on the sidelines of a biotech conference in London.
“Traditional medicine in some ways rings alarm bells because it’s esoteric. But by applying data-driven approaches, we can read through the noise and what is hogwash versus what is a legitimate potential plant-disease association,” said Ms Chandra.
“What we find is that when you look at all the drugs on the market today, a large proportion of them were discovered this way.”
Mr Sunak this week visited Moorfields Eye Hospital in London where he was shown a retinal scan system that can detect signs of blindness, heart attacks, stroke and Parkinson’s disease after being trained on data from millions of patients.
Machine translation has been around for years but the output is a lot better than it used to be.
Google says advances in AI make it possible to upload an image of a sign and have the text translated automatically.
Experts have raised concerns over the software’s use in sensitive settings such as immigration appeals, especially where less widely spoken languages are involved. There are fears that officials are cutting costs by asking people to rely on translations that are by no means foolproof.
Meanwhile, the language learning app Duolingo has turned to AI to think up sample sentences. Humans dictate the difficulty level, the theme and the grammar point to be tested, and AI does the rest.
London’s Metropolitan Police has confirmed it uses AI as part of a facial recognition system, despite concerns that the technology struggles with non-white faces and can wrongly identify people.
The “digital twin” concept has also been applied to detective work. Ukraine has acquired hand-held scanners to create 3D models of crime scenes, preserving what could be evidence of war crimes.
“We hear now about a lot of great achievements of AI,” said Artyom Yukhin, the president of Artec, the company that made the scanners. “And still, there are not that many achievements in applying neural networks to 3D data. So we are working in this sweet spot.”
The technology has been used by Dutch authorities who investigated the MH17 plane crash in 2014 and built a piece-by-piece model of the downed Boeing jet, Mr Yukhin said.
The Dutch customers “explained that when they document the crime scene, it’s quite different from Sherlock Holmes’s times”, he said. “Sherlock Holmes would understand what happened on the spot and then it’s done.
“Normally, everything does not happen on the spot. [Investigators] shouldn’t think about what is more important and less important and be biased. They need to document everything with good quality and then to have the digital replica of it.”
Mr Sunak described an increase in government efficiency thanks to an AI doing paperwork in the benefits office. While humans could spend a week producing 11 bundles of papers for tribunals, an AI can do it in less than an hour, he said.
Germany, which is notorious for a pen-and-paper bureaucracy that still uses fax machines, has similarly called in AI bosses in a bid to speed things up.
Archives are also pondering whether AI can help write the first draft of history by sifting through a mountain of official documents. US officials are piloting a system to automatically generate answers to freedom of information requests.
Netflix says it invests heavily in machine learning to help personalise TV and movie recommendations and see the features that make particular shows popular.
Spotify this year launched an “AI DJ” that “knows you and your music taste so well that it can choose what to play for you”.
And YouTube is using AI to “reshape advertising” around videos, asking it to automatically mark a brand’s homework against criteria such as the prominence of its logo.
AI is being used in schools to set questions, outline lesson plans and help students check their work, as The National recently reported. Some exam boards are open to software such as ChatGPT being used in tests.
“In some contexts, it may be perfectly valid to allow the use of AI tools, just as some assessments allow candidates to use calculators, or search engines,” said Alex Scharaschkin, executive director of research and innovation at exam board AQA.
A second board, OCR, says it will be acceptable for students to use AI for initial research from next month. They can also use a few sentences of AI-generated text if they comment critically on its content.
Britain will use the November 1-2 summit at Bletchley Park to seek consensus about AI’s risks, with officials expecting its capabilities to rise further in the next two years.
The world will be invited to commission a “state of AI” report from a group of experts modelled on the Intergovernmental Panel on Climate Change, which gives the definitive verdict on global warming.
Mr Yukhin, the 3D scanner maker, expects to see “a lot of cool results in the coming years” because AI is “changing the game a lot”. At the same time, Mr Sunak has committed himself to clamping down on its risks.
Gavin Poole, who runs a tech campus in London called Here East, said AI had become a “focal point of the international policy agenda”.
It is going to remain there as we grapple with how to restrain its power, while also facilitating safe innovation and growth, he said.
Mr Sunak has been gently mocked for his interest in AI, a subject that rarely appears in a list of top voter concerns. As it makes its presence felt in everyday life, that could change.