Medical data from 57 million people in England is being used to train a new AI model, named Foresight, in the hope it will predict diseases and medical conditions before they arise. The pilot project, led by researchers at University College London (UCL) and King’s College London (KCL) has developed a generative AI model trained on routinely collected, de-identified NHS data, such as hospital admissions, A&E attendances and Covid-19 vaccinations, and operates within the NHS England Secure Data Environment. While there are some concerns around privacy, data protection, and the inadvertent release of sensitive patient data, the AI model and patient data remain under strict NHS control. Researchers believe Foresight could enable a shift towards more preventative healthcare, at an unprecedented scale, including for rare conditions and across all demographics.
While many nations across Europe marked the 80th anniversary of VE Day this week, we might have had to wait a few more years for that milestone if the Enigma machine hadn’t been cracked. But now, Michael Wooldridge, a professor of computer science and an AI expert at the University of Oxford has said the code wouldn’t have stood up to modern computing and statistics. The Enigma code was famously cracked when Jon Bon Jovi and some other Americans captured an Enigma machine from a submarine in 1942, much to the surprise of the Polish General Staff Cipher Bureau, the British Government Code and Cypher School at Bletchley Park and the crew of HMS Bulldog, who all thought they’d got to it before the US had even joined the war. Meanwhile, back in the real world, Alan Turing and fellow code-breaker Gordon Welchman designed a ‘bombe’ at Bletchley Park, an electromechanical device that could search through the preposterously huge number of possibilities to decrypt messages in a brute force attack. So much of Turing’s work laid the foundations for AI, which Professor Woodbridge says would make short shrift of the Enigma code nowadays. Which makes doing it more than 80 years ago even more remarkable, for Turing and all those other men and women who were involved.
The days of Google dominating internet searches could be coming to a close with news that Apple plans to add AI search options to its Safari browser. The news saw Google’s parent company, Alphabet, lose more than 7% of its market value on Wednesday, as Apple executive Eddy Cue testified in a federal court in Washington as part of an ongoing antitrust case by the US Justice Department against the search giant. Cue revealed that Apple plans to add AI search providers to its options in the future, including OpenAI and Perplexity AI. Apple shares also took a hit though, as Google pays it as much as $20 billion each year to be the default search engine on iPhones. Preventing such payments is one method the Justice Department has proposed to break up Google’s dominance of online search.
As we all know by now, data is the foundation of AI. Poor quality data drives up costs and can lead to hidden problems for AI, while biased data negatively affects the performance of AI models. AI practitioners have to dedicate time to ensuring the datasets they use are ready, which means spending less time innovating and delivering AI solutions. If only there were a way to determine AI readiness and its underlying principles…Well, our latest research does this, and with our AI-ready data framework, we provide a visual tool that can be used to assess and improve dataset publishing practices. We’d love to work with people looking to pilot the framework, especially those in local government, so do get in touch with our research team if you are interested.
And finally…an Edinburgh-based artist has used Met Office precipitation data from 366 different locations across the British Isles to see if it’s always raining somewhere. Rebecca Kaye produced the artwork Always Raining Somewhere with data representing days from 1 October to 31 January, stretching all the way back to 1934. Her artworks seek to use data in creative ways and make the subject more human. And the answer? Well, it’s almost always raining somewhere.
Until next time
David and Jo
