AI Briefing: A new life for your old smartphone photos
I’m not a mind-reader, but I can tell there’s an old mobile phone buried somewhere in your house.
This phone no doubt holds treasured but very grainy photos from a time before camera phones were worth using – and before it was easy to transfer them to a computer.
Thanks to AI, it might finally be time to dust off that Nokia handset and, more importantly, the digital snaps within it.
A new research paper details how AI, trained using 50,000 pictures, can estimate how your photo should look behind all that ‘noise’ or ‘grain’.
The research was conducted by Nvidia, MIT and Aalto University and could one day be part of the photography process.
More importantly, it’s a chance to clean up old snaps of your friends in trucker caps, studded belts and baggy jeans.
Henry Kissinger worried about AI overlords
Controversial ex-politician Henry Kissinger, who served as national security adviser and secretary of state to Presidents Nixon and Ford, doesn’t think the world is ready for AI.
In a post for The Atlantic, Kissinger described his recent exploration of the technology, and has come to the conclusion that it poses some tough questions for our global society.
“Science fiction has imagined scenarios of AI turning on its creators,” he wrote. “More likely is the danger that AI will misinterpret human instructions due to its inherent lack of context.”
Kissinger urged the US government to consider AI a ‘high national priority’ and should suggested that it ‘consider a presidential commission of eminent thinkers’ on the topic.
First Minister of Scotland Nicola Sturgeon shared the article on Twitter, adding: “If you want something non-party political to get your intellectual juices flowing, I recommend this by Henry Kissinger.”
Not content with one spot on this this week’s AI round-up, the Massachusetts Institute of Technology has released a second, equally impressive new tool for scrubbing up old content.
Its PixelPlayer has been trained on just 60 hours of video and with the power of AI can now single out multiple sounds from one video.
Whilst the real-world applications aren’t very exciting yet (how many musical duets have you watched recently?) the team behind the project hope that one day the technology will be able to separate any performance.
One day we might be able to turn up Jimi Hendrix’s guitar – or perhaps turn down Morrissey’s vocals.
Creeped out? We can already tell
No technology combines feelings of amazement and horror quite as well as artificial intelligence.
For every jaw-dropping new invention, it seems there’s another which has us wishing for a power outage.
A new consumer survey asked 2,577 people from the UK, France, and Germany how they feel about AI. When asked which technology they thought was ‘creepy’, AI, marketing automation and ‘emotion detection technology’ topped the list.
Over half of those surveyed thought that technology which adapts to your emotions – mostly to sell you things – was creepy, not cool.
To find out which tech topped the cool list, read more here.
A positive drugs test
A team from Stanford University is using AI to better predict the potential side effects from combining medication.
The technology is sorely needed. According to the university, there are about 1,000 known side-effects from taking any one medication. With 5,000 drugs on US the market that makes for nearly 125 billion possible side effects between all possible pairs of drugs.
With so many possibilities, it’s not surprising that doctors often can’t predict what side effects might come from a cocktail of medications taken at the same time.
Enter artificial intelligence. Rather than trying to categorise each combination, the research team which lead this new technology studied how drugs affect the underlying cellular machinery in our body.
The mammoth research project included studying how 19,000 proteins in our bodies interact with each other and how different drugs affect them.
Using more than four million known associations between drugs and side effects, the team began to identify patterns in side-effects, based on how drugs target different proteins.
“Today, drug side effects are discovered essentially by accident,” said Jure Leskovec, an associate professor of computer science at the university.
“Our approach has the potential to lead to more effective and safer healthcare.”