Artificial intelligence and machine learning are advancing at a breakneck pace. So quickly, in fact, that it’s striking to recall that just ten years ago the AlexNet model was winning ImageNet and helping launch the shift that turned deep learning into a true technology movement. Now, after years of coverage focused on games, we’re seeing more and more innovation aimed at the practical world.
Over just the past couple of years, AI/ML systems such as GPT-3 and AlphaFold unlocked abilities that helped spark new products and companies, and expanded our sense of what computers are capable of.
With that context, we wanted to look back at our AI/ML coverage in Future from the first half of the year, and also bring you up to speed on some — though definitely not all — of the big industry changes during that period. As you’ll notice, a mix of large language models, generative models, and foundation models is drawing a great deal of attention, and we are only beginning to grasp what they can do and how organizations beyond large research labs can make use of their power.
The Future Focus: How to Harness Rapid New AI/ML Advances Today
How to Apply Huge AI Models (Like GPT-3) in Your Startup by Elliot Turner / Hyperia Labs
AlphaFold, GPT-3 and How to Augment Intelligence with AI by Niko Grupen / Cornell
AlphaFold, GPT-3 and How to Augment Intelligence with AI (Pt. 2) by Niko Grupen / Cornell
Data50: The World’s Leading Data Startups by Jennifer Li, Sarah Wang, and Jamie Sullivan / a16z
New Architectures for Modern Data Infrastructure by Matt Bornstein, Jennifer Li, and Martin Casado / a16z
A Decade of Deep Learning: How the AI Startup Experience Has Changed with Richard Socher (Q&A) / you.com over time
7 Techniques for Building Reliable AI Models by Beena Ammanath (book excerpt) / Deloitte
The Two Things We’ll Need for the Next AlphaFold with Daphne Koller (Q&A) / Insitro
Industry Focus: Images, Words, and More Coding
Competitive Programming with AlphaCode / Deepmind
Teaching AI to Translate Hundreds of Spoken and Written Languages in Real Time / Meta AI
Pathways Language Model (PaLM): Scaling to 540 Billion Parameters for Breakthrough Results / Google Research
DALL-E 2 / OpenAI
Imagen: Text-to-Image Diffusion Models / Google Research Team
These kinds of advances, along with the growing understanding of how to apply them, are why we’re committed to expanding our AI/ML coverage and, especially, to how we’ll see it used in real-world environments over the next few years. From biotechnology to television, we’re on the verge of a major rethink of what’s possible and how software can help people realize their boldest ideas. If you’re building something exciting and new in the AI/ML space and want to share your views on where we’re going, please send us a pitch.