Last week, the iMerit team was fortunate enough to sponsor and attend the O’Reilly Artificial Intelligence Conference in San Francisco. The sold-out event brought together business leaders and AI innovators from across the industry to discuss where AI is today and where applied AI is headed.
This conference was one of the highlights of our year so far. It stands out for the lineup of speakers, great vendors, and engaged audience from across all fields, from logistics and manufacturing to health and media. The conference featured cutting-edge science and business implementation, focusing on topics like increased AI accessibility, innovations in AI techniques, and how AI is driving a paradigm shift in computing itself.
Here is a bit of a recap of what we took away from this very informative and innovative conference.
Trends in Artificial Intelligence
Andrew Ng, Co-Founder of Coursera, an online learning platform, and one of the most sought-after innovators in Computer Vision, gave a keynote whiteboard session (the first I’ve seen on the big stage) for his talk entitled “AI is the New Electricity” because the invention of electricity affected all industries and so will AI. In his talk, he discussed what AI does, AI products, and how the AI era is changing the basis in competition today.
Andrew went over some of the innovations in AI happening today such as speech recognition is search (Alexa), input text and output audio (Baidu), etc. and highlighted the fact that the weakness to all of these innovative algorithms is their hunger for structured data.
He addressed a question that he often gets: “Neural networks have been around for decades, why are they just now taking off?” Because of digitization, we have access to more and more data. What became apparent about five years ago is that there is a direct correlation between performance and the amount of data you use to train neural networks.
He continued to discuss the economic opportunity of supervised learning and what comes next, including transfer, unsupervised and reinforced learning. You can see highlights from his keynote here.
AI in Healthcare
There were several sessions dedicated to how AI is working in the healthcare industry including new drug discovery and disease detection.
Blake Borgeson, Co-Founder and CTO of Recursion Pharma talked about how his company is revolutionizing the way we discover new drugs while Nan Li, Principal at Obvious Venture highlighted the technical and philosophical shift happening and the fundamental un-doing of the way we think about problems today.
Blake’s company is using AI to leverage a brute force method of discovery. This dramatically changes how things are done today. To bring a new drug to market, it currently costs between $2-5 Billion and can take over ten years. Not only that, because of the way the system is built, they may be doubling down on the wrong candidate.
By using computer vision technology to identify the ‘health’ of cells in an image, Blake’s team can quickly and efficiently conduct experiments that are rooted in deductive reasoning (100,000’s experiments per week at $0.25 each). Here is a recreation of the chart Blake shared to demonstrate how Recursion is changing the way we look at drug discovery today through a shift from empiricism to understanding – instead of asking over and over, have the machine learn over time.
Cloud AI
Google’s Philippe Pouponnent focused on the democratization of AI, what Google and others are doing with AI and how fast it is being adopted. Showing user data, Pouponnet demonstrated the hockey stick graphs showing adoption of Google’s AI products, including Google’s visual translation product.
Philippe shared a case study working with Airbus. They have been taking satellite photos from around the globe and cleaning them up with human in the loop labeling. An example of a problem they have been having for over a decade is computer being able to tell the difference between clouds and snow. Over time, and with a lot of labeled data, they have been able to dramatically improve error rate and development time.
For more, watch Philippe’s full presentation.
Is AI Evil?
Tim O’Reilly took the stage for final remarks. His talk on ‘Our Skynet Moment’ opened with Elon Musk’s fear for that AI is the most serious threat to the survival of the human race. Relating our connected world to that of Skynet from Terminator, we are defining a period of the struggle for human freedom.
Our world, from getting from A to B to world business markets are becoming increasingly managed by algorithms. Tim urges that we must rewrite current algorithms that shape our economy if we would like to see a more human-centered future. This all leads to O’Reilly’s new book, The Future And White It’s Up To Us, where he discusses topics such as the decline of jobs with the rise of technology. You can check his new book out here.
For more of the excellent sessions put on by today’s AI innovators, check out the O’Reilly blog and be sure to check out their upcoming events in 2018.
