Today we discuss the mystery behind Cerebras, a stealthy startup funded by Benchmark between $25 and $30 million to develop deep learning hardware, the last report
released by the White House on the effect of new AI technologies on the economy, and the solution to understanding how AI algorithms work.
The News Team
Follow us on SoundCloud to be updated on our podcasts.
Smile Vector is a Twitter bot that scrapes the web for pictures of faces, and then it morphs their expressions using a deep-learning-powered neural network. Its results aren’t perfect, but they’re created completely automatically, and it’s just a small hint of what’s to come as AI opens a new world of image, audio, and video fakery. Imagine a version of Photoshop that can edit an image as easily as you can edit a Word document — will we ever trust our own eyes again?
The newly released White House report states AI will lead to some jobs lost, but we still need to invest on it. Jason Furman, the chairman of the Council of Economic Advisers, said in a call with reporters: “As we look at AI, our biggest economic concern is that we won’t have enough of it, that we won’t have enough productivity growth.” The report calls for further investment in AI research and development. Specifically, the White House sees the technology’s applications in cyber defence and fraud detection as particularly promising.
AI algorithms are usually only programmed to provide an answer based on the data they’ve learned. We can see their conclusions, but most of the time we don’t know how they arrived at them. That limits our ability to improve AI when something goes wrong, but now, a research on attentive explanations is set to change the game.
Benchmark is doing another big investment before the end of the year, and this time it’s in a hardware startup called Cerebras Systems. Details on the company are extremely sparse though: the company is working on a specialized next-generation chip (possibly a GPU) for deep-learning applications. The company’s description from LinkedIn (really the only thing that can be found from a Google search) says: “Cerebras is a stealth-mode startup backed by premier venture capitalists and industry leading technologists. We are serially successful entrepreneurs dedicated to solving problems others are afraid to tackle.”
This podcast presents an entertaining historical perspective on the evolution of the neural network concept from its biological origins and provides some examples of recent successes in neural network machine learning algorithms.
Honda’s 2017 CES exhibit will include the NeuV, a concept autonomous EV commuter vehicle equipped with AI. “From reducing traffic congestion to creating new modes of in-car connectivity, visitors will have an opportunity to explore and demo technologies with the potential to make people’s lives better. The exhibit will include the NeuV, a concept automated EV commuter vehicle equipped with AI called ’emotion engine’ that creates new possibilities for human interaction and new value for customers. Continuing its pursuit of open
innovation and collaboration, Honda also will announce initiatives with startup companies and global brands that will create a more productive and enjoyable mobility experience.”
Noom’s platform combines human coaching and AI to deliver scalable behaviour change programs that prevent and manage chronic conditions. The company’s current programs target cardio-metabolic conditions like obesity, pre-diabetes, diabetes, hypertension that can be mitigated through better nutrition and exercise. Participants follow disease-specific curriculums that include daily tasks assigned by a trained coach, short-form educational content, food-logging, exercise-logging, one-on-one coaching and group support. The British Medical Journal Open Diabetes Research & Care recently published an article about the efficacy of Noom’s Diabetes Prevention Program (DPP); it was the first-ever peer-reviewed study of a fully mobile DPP. The study found 64% of participants who completed Noom’s program lost over 5% of their body weight. Results were comparable to the CDC findings from the traditional in person diabetes prevention program and far more successful than other virtual DPP providers.