This week in AI #2: This is no fake news, or is it?
We’re back this week with the most thought-provoking articles we could find: two applications of DeepLearning with awe-inspiring results, big players making their move and Elon being Elon.
1- This one speaks to our heart…
Earlier this year, Standford researchers built an algorithm to detect skin cancer that performed on par with dermatologists diagnosis. Well, they’ve learned their lesson and they’re back at it! This time they’re trying to detect abnormal heart rhythm, quite literally when your heart skips a bit. The big difference being that their diagnosis is not just on par with the cardiologists board but, lo and behold… completely outperforms it! You can read the full article here.
2- Microsoft to focus on general-purpose AI
With DeepMind opening its new AI center in Edmonton last week, just after the Montreal-based startup Element.ai announced it raised $102M , it’s now time for Microsoft to join in on the fun. The tech Giant now officially announced its Research AI team. The goal is simple: advance the AI frontier but more importantly distill it in all of Microsoft products which is where the real challenge is!
3- Google’s new fund
Last week was full of announcements! It’s been rumored for some time now but Google finally revealed its new Gradient Ventures fund that will focus on early AI startups. Apart from funding they say they will connect the startups with ‘Google’s resources, innovation, and technical leadership in artificial intelligence’. Lately Google has been doing its best to push the AI community to adopt its Machine Learning tools, especially TensorFlow. Which raises the question: what better way to promote them than nurturing the next tech unicorn into using them internally from the beginning?
4- Fake news just got way more real
Recently it’s been difficult to tell facts from fiction in heated debates, but sometimes a video clip comes up which ends the discussion once and for all. Well, that was way too easy for some University of Washington researchers! Using existing footage, they trained an algorithm to generate a real-looking video speech just from any audio recording. You might think this would be reserved to some over-exposed celebrity but the dataset used was composed of only 17 hours of Obama footage; at a time when everybody is sharing videos, sometimes several times a day, this might not be that much of an issue. Head over to their paper if you want to know all the technicalities!
5- Elon Musk weighing in on regulations
It’s been clear from the start: Elon has plans for us. And by us I mean mankind. Whether it be leaving the coal-and-oil era to go clean with Tesla and SolarCity or colonizing Mars in case something goes wrong on Earth; he’s basically been trying to make us transition from a type I to type II civilization. He’s also very concerned with AI: with Neuralink, his new company, he aims at allowing us to communicate with the next strong-AI… you know, just so it doesn’t think of us the way we think of ants. Well Elon just gave a public talk to a bunch of regulators at the National Governors Association; the undertone being that it might be a good idea to talk about where we’re going before flooring the gas pedal of our brand new Tesla Model 3.
What do you think about it? Should we live and learn or pause and make sure we’re not heading for the wall? Let us know in the comment section, in the mean time we’ll be implementing all the latest tips and tricks of deep learning so that you can focus on your expertise and leave the rest to us!
PS: If you want to know more about what Elon is really trying to achieve for humanity then waitbutwhy has got you covered, just don’t plan on doing something for the next few hours.