Happy Sunday and welcome to the latest edition of Inside AI! For those of you who are new, I'm Rob May, CEO of Talla and active investor in the A.I. space. If you have an early stage A.I. startup looking for investment, I'd love to hear from you.
This week's issue is awesome. If you agree, I hope you will forward this to a friend so they can subscribe too. Or if you really want to support us, go Premium.
One correction from Wednesday's premium edition: The Wired article I linked to mentioned that Yoshua Bengio had dropped his goal of staying neutral and signed on to support Microsoft, but on his Facebook page, Bengio says that is an error and that he is an advisor to multiple companies.
Also, we hoped to have an interview about Workbench's new report on the future of enterprise tech, and the machine intelligence insights in that report. The launch was delayed so, we will cover that next week.
-- Big Idea --
The most interesting thing I read this week was this article about Silicon Valley's new obsession with brain tech. Neurotechnology is important because many people believe it is the way we augment ourselves to keep pace with the machines as they get smarter. From the article: "The typical Silicon Valley attitude is that if you throw enough money at something, you can solve the problem,” Sajda says. "While that approach may work for applied technology, he says, it doesn’t necessarily work if there are fundamental science questions that need to be answered—and there are many unanswered questions in neuroscience." There are lots of reasons to be skeptical, but, on the flip side, it is often the suspension of disbelief that enables new breakthroughs in science and technology. We will continue to follow neurotechnology but for now, I come down on the side that it's probably further off than we think.
-- MUST READ LINKS --
Cisco Unveils Subscription Service For A.I. Based Networking. Venturebeat.
This is really interesting because it's an early example of A.I. moving to a new part of the technology stack that hasn't seen much A.I. adoption before.
The Inside Story of Talla's $8.3M Series A. Medium
This is my post about the round my company recently raised. It has some good insights for raising for A.I. companies, and how you may approach VCs.
Generative Adversarial Networks: History and Overview. Towards Data Science.
This is a highly technical piece, so not for everyone, but GANs are an important innovation in A.I., so it is worth investing the time to understand them at a high level.
Where A.I. Will Pay Off The Most In Health Care. Fortune.
A summary of Accenture's new report on health care and A.I. Lots of potential and lots of investment going into the space.
Learning About The World Through Video. TwentyBn Blog.
This blog post is a great look at the importance of A.I. and video going forward, and includes an announcement of new datasets the company is releasing to help train A.I.s.
-- INDUSTRY LINKS --
The Man Who Helped Turn Toronto Into A High Tech Hotbed. NY Times.
Jack Ma Warns About The Dangers of Artificial Intelligence. RT.
A Brief History of Neural Network Architectures. Topbots.
Redefining Basketball Positions With Unsupervised Learning. Medium.
One Machine Learning Model To Rule Them All. ZDNet.
How I Built An A.I. To Sort 2 Tons of Lego Pieces. IEEE Spectrum.
In The A.I. Age, "Being Smart" Will Mean Something Completely Different. Harvard Business Review.
Google Launches An A.I. Powered Job Search Engine. Techcrunch.
The Cross Office Machine Intelligence Revolution. HedgeWeek.
Why A.I. Works. Artificial-Understanding.
When Agile Meets Machine Learning. Towards Data Science.
The World's Most Powerful Particle Collider Uses A.I. To Expose Hack Attacks. Scientific American.
Tesla's New A.I. Guru Could Help Cars Teach Themselves. MIT Tech Review.
Snips Raises $13M For A Voice Platform Alternative To Google and Amazon. Venturebeat.
Google's New A.I. Fund Makes Their First Investment in Algorithmia. Techcrunch.
How Faulty Data Breaks Your Machine Learning Process. Dataconomy.
3 Key Trends Shaping The Data Science Hiring Market. KD Nuggets.
A Review of "Deep Thinking", Garry Kasparov's Book About His Battle With Deep Blue. Guardian.
Intel's Future Lies With Artificial Intelligence. Nasdaq.
-- Research Links --
Unsure When To Stop? Ask Your Semantic Neighbors. Link.
Towards Deep Learning Models Resistant To Adversarial Attacks. Link.
Sparse Neural Network Topologies. Link.
Many thanks to Inside AI's corporate supporters
-- Commentary --
Generative Adversarial Networks (GANs) may take us into a "post truth" world. While pundits have already been using the term "post truth" to talk about fake news and such, the technology to really create a post truth world is here. This video of a horse transformed into a zebra from a tool called CycleGAN is one of the most remarkable things I've seen. Imagine what can happen in a world where putting your face and voice on other pictures or videos is easy and fast.
Some of it will be good. You can imagine how cool it would be to see your favorite band joined by your other favorite band on stage at a famous historic concert. You can imagine having your favorite actor also be your favorite teacher, and the impact on learning if you suddenly show a GAN generated LeBron James delivering a lecture on math to young kids.
But I worry a technology like this will see more nefarious uses. You can see political ads where someone has a video of a supporter of a specific position suddenly voicing the counter position. You can see political scandals where the face of any congressperson is dropped into a pornographic movie and released publicly. You can see a video call from your college aged daughter, who is traveling abroad, saying they've been mugged and now you need to wire money to help them out, only to realize hours later that your daugther was fine, and a scammer replicated her face and voice to con you. In this world, it might become very very tricky to know what is true.
How do we fix this? I think we will see some sort of A.I. driven blockchain for media. If you haven't followed bitcoin, or some other blockchain technology, it's the perfect solution for something like this. Keeping a distributed ledger that requires several independent parties to verify could identify edits and changes, particularly those A.I. related, and could help to solve this problem. A.I. and blockchain together is something I'll explore in coming newsletters, as issues like this have me thinking more about it.
It's early days of this tech, but by 2018 I bet you start seeing your first stories of people duped by criminals using these techniques. Hopefully solutions, be they blockchain based or otherwise, aren't too far behind.
That's all for this week. Thanks again for reading. Please send me any articles you find that you think should be included in future newsletters. I can't respond to everything, but I do read it all and I appreciate the feedback. And thanks to Kylie.ai for being a sponsor.
-- ABOUT ME --
For new readers, I'm the co-founder and CEO of Talla, I'm also an active angel investor in A.I. I live in Boston, but spend significant time in the Bay Area, and New York.