What Are The Most Promising Areas Of Machine Learning Research Right Now?

These questions originally appeared on Quora: the place to gain and share knowledge, empowering people to learn from others and better understand the world.

Answers by Sam Altman, Y Combinator and OpenAI, on Quora:

What are the most promising areas of machine learning research right now?

The most important meta-learning of my career so far is that scale is always more powerful than I think. I’ve seen this for individual companies (networks effects, economics of scale, etc.), the power of the YC network because of its scale, and in what happens when you scale up machine learning models.

I think one of the most promising directions right now in ML research is figuring out how to scale up ideas that are already working. And it’s the kind of thing that doesn’t sound as exciting as a brilliant new algorithm, and it’s hard, so it’s tempting to look for another answer.

I am particularly excited about progress in unsupervised learning, and in combining UL models with goal-directed agents.

What is the goal of OpenAI?

The goal of OpenAI is to ensure that artificial general intelligence (AGI) benefits all of humanity. We want to ensure that the field builds a good future—one where the benefits of AGI are widely distributed and used for individual empowerment, we are able to solve problems that are otherwise too difficult (like stopping climate change, making education great worldwide, and curing many diseases in a short period of time), and that the systems we develop are safe.

I believe that if AGI truly does get built, it will be the most important technological achievement about human history, and that life on the other side will be very different. The magnitude of this change is so big that we designed a new corporate structure to let us balance our capital needs with our mission.

What are your three main criteria you look at when investing in a startup?

1. Is the mission inspiring? (Without this, it is difficult to concentrate enough talent over time, and hard for me personally to keep focused on it.)
2. Is the market going to grow quickly? (I don’t care too much about the current TAM but I care a lot about it what it will be in 10 years. I also try to differentiate between real trends and fake trends—eg, are initial users of a platform using it every day?)
3. Would I work for the founders?
4. Does the strategy make sense (e.g., is there a network effect that will get built up over time, is the business model sensible, etc.) and is the product so good that people are telling their friends about it?

These questions originally appeared on Quora – the place to gain and share knowledge, empowering people to learn from others and better understand the world. You can follow Quora on Twitter, Facebook, and Google+. More questions: