Mustafa Suleyman has been working in artificial intelligence for 12 years, trying to figure out how to use machine learning systems and AI to do important things in the work and have impact at scale.
“And over the years, I've been lucky enough to be at the forefront of a lot of cutting-edge applications of AI,” he told TechCrunch. “Over the years, that experience has given me a really good intuition, for when a piece of AI is ready for the real world, and when it's not. The projects that I've seen fail are mostly because people overestimate how good AI is. People think that AI is this silver bullet and it can solve all your problems, but actually you have to craft the environment and the application problem correctly.”
Suleyman is now putting that intuition to good use on the venture capital side. After previously investing in companies alongside Greylock partner Reid Hoffman, Suleyman made the full leap to join Greylock as a venture partner.
He joins the firm from Google, where he was vice president of AI product management and AI policy. Prior to that, he co-founded and led DeepMind, an AI company acquired by Google in 2014.
"AI will no doubt touch every aspect of our lives in the coming years, and we at Greylock believe there's an abundance of opportunity for entrepreneurs to continue building in this space.” Hoffman said via email. “Mustafa is visionary, knowledgeable and connected across the vast AI landscape, and we know he'll be a valuable resource to our existing portfolio, and a board member of choice for new AI investments."
Suleyman is eager to work with early-stage founders, who he said are “super fearless and really energetic people” who aren’t afraid to take a risk when they see opportunity.
He believes there is much opportunity in the AI companies that are out there. He says the time has come for AI due to the ecosystem maturing, and more people understand the strengths and limitations of AI and how to use it. AI is also becoming more accessible and usable by people who don’t necessarily have a technical background, but are creating new ways to use machine learning.
As such, he says that AI is at an inflection point: We now have AI systems that can generate new text, conversational sentences and whole paragraphs, which is approaching human-level performance.
“The range of things that entrepreneurs can do when you can basically have an API that can talk to your users in natural language is amazing and the imagination is the limit,” Suleyman added. “Combined with the explosion of the metaverse and all the excitement around that in the last couple years, I definitely think that AI has a central role to play in virtual reality, in gaming and the metaverse.”
While those are a few areas where he sees AI winning, there are some areas where he thinks AI is not quite there yet, including large-scale infrastructure, manufacturing and logistics distribution, and he is looking for those companies stepping up to build AI into scheduling and coordination of shipping, shipment tracking and route optimization.
When looking at what is down the pipeline for AI, he again sees the metaverse and gaming dominating the space as characters and avatars come to life -- think “Ready Player One,” where there is an animated parallel world in tandem with ours.
“It'll be able to do planning and prediction, so it won't just be kind of generating scripted or written responses, it will actually be emergent and responsive to the environment,” Suleyman said. “On the enterprise side, the time for AI and medical imaging is definitely now that we've proven it works in research, and now we're ready for large-scale production that's going to be very successful.”