Last Week in AI
Every week, my team at Invector Labs publishes a newsletter to track the most recent developments in AI research and technology. You can find this week’s issue below. You can sign up for it below. Please do so, our guys worked really hard on this:
From the Editor
Can AI agents learn human values? That question has hunted researchers since the early days of AI. For the most part, we know very little about how values such as justice or compassion materialized in the human brain. What we do know is that most of those values are the result of social interactions rather than pre-wired connections. Following that line of thinking, could we leverage principles of social sciences like phycology or sociology to align AI with human values? Researchers from OpenAI seem to think so.
This week, OpenAI published a detailed research paper exploring the importance of social sciences in the future of AI. The paper incudes an experiment in which an AI agent learns to identify human values by following a debate game between humans. Debating is an activity that requires both intelligence and social skills and provides a unique setting to explore human values. The OpenAI study reflects that the future of AI is not only in the hands of computer scientists and mathematicians but requires a lot of help from social scientists.
Now let’s take a look at the core developments in AI research and technology this week:
Research
OpenAI researchers published a paper about the important of social sciences to create AI models aligned human values.
>Read more in this blog post from OpenAI
Microsoft announced the winners of the multi-agent reinforcement learning challenge.
>Read more in this blog post from Microsoft Research
Deep learning legend and Facebook’s Chief AI Scientist Yann LeCun gave en enlightening talk about the future of AI hardware.
>Read more in this blog post from Facebook Research
Cool Tech Releases
Horovod, the training library created by Uber AI Labs, added support for PySpark and Apache MxNet.
>Read more in this blog post from the Uber engineering team
Staying with Uber, Pyro was accepted as an incubation project by the Linux Foundation Deep Learning Technical Board.
>Read more in this blog post from the Uber engineering team
Microsoft released a new version of SEAL, its homomorphic envryption library that is considered one of the pillars of decentralized AI.
>Read more in this blog post from Microsoft Research
AI in the Real World
Wikipedia published the first timeline about the evolution of AI
>See the whole timeline in the new Wikipedia page
Top investors are relying on AI for gaining an edge in the market
>Read more in this coverage from CNN
Students from the Massachusetts Institute of Technology(MIT) are working on diverse projects to advance research in AI.