Deepfake vs AI; AI vs regulation [Wired]
Deepfake videos are becoming so sophisticated that even AI has trouble keeping up. In a bid to thwart the potential use of the technology for political sabotage in the 2020 US presidential race, Google released last week several thousand videos for research and machine-learning. In parallel, Facebook and Microsoft are building their own dataset for release in December. But the face-swapping technology is evolving so fast that the huge database of videos collected may already be outdated. For Facebook and Google, the race is on to fix the problem before policymakers hit back with regulation.
Think again and meet SenSat, the UK company which will make infrastructure planners’ – and maybe video game developer’s – life easier. SenSat uses AI to create digital representations of real-world locations. By creating “digital twins” of the real world, SenSat enables industries to make better informed decisions about big projects by improving safety, cost-efficiency, waste generation and collaboration, and by reducing the carbon generated in a project. The firm has just raised a $10 million funding round led by Chinese internet giant Tencent, who sees in SenSat no more than the solution to “mass digital automation of offline industries.”
The modern Lassie is here [Wired]
Soon, your dogs might be able to communicate with you! Well, service animals, using a special vest – but still pretty cool. The FIDO project at Georgia Tech in the US is developing special vests for service dogs that have pre-recorded messages in them like “My owner needs your attention! Please follow me!” The message is activated by the dog pressing a button on the vest. While there is still a ways to go before our dogs can talk back to us – hopefully this tech innovation will help make it safer for those who use service animals.
The Facebook election drama continues [Financial Times]
After it became apparent that Facebook was seriously impacting elections around the world, the platform said it would hand over a petabyte (a million gigabytes) of user data to academics so they could study Facebook’s impact. Well that petabyte was never delivered to researchers, with Facebook citing issues like privacy. Several funders might end their involvement with the project. One issue is GDPR – which has apparently made Facebook nervous about sharing user data, even when it can’t be attributed to a specific user. Facebook said the research would continue, but it may be a while before we see any results.
In case you haven’t had enough…