Cointelegraph’s recent Ask Me Anything session on AI and blockchain brought together Foresight Ventures partner Tony Cheng, Hack VC managing director Ed Roman and Dragonfly Capital managing partner Haseeb Qureshi.
The speakers discussed possible AI use cases in the crypto space, prominent startups that are already using AI, and investor interest in similar businesses. All guests noted that it’s too early to predict the evolution of Web3 and AI collaboration or the appearance of any unicorns at their intersection, as large language models are just emerging. They also emphasized the very different nature of these technologies, with AI being authoritarian while blockchain and Web3 remain democratic, non-hierarchical and decentralized.
“We get pitches that are mostly solutions looking for problems. For example, some propose to build infrastructure to serve use cases, but the problem is that those use cases don’t exist yet,” noted Roman, speaking about AI startups. Cheng agreed with him: “I think it’s still quite early for AI investments, especially with Web3’s perspective. “We’re looking at different ideas from startups like ZK, machine learning and on-chain AI, but we haven’t seen any real use cases yet.
“When people talk about AI products, they mean products with AI features. But you cannot build a startup with only AI as a feature. Super advanced models have not appeared yet. So what you’re left with is to analyze what your competitors are doing best, what the audience is interested in, use AI models like OpenAI and build on that,” Qureshi added.
Talking about potential AI use cases for blockchain, the speakers made several suggestions, recommending similar services based on the history of user actions in DApps, detecting fraud and smart code anomalies, and AI acting similar to a kind of financial advisor. In addition, they suggested improving the user experience, with AI explaining the inner mechanics of smart contract and DApp inner actions which could potentially help crypto mass adoption.
“The main role of AI is not to invent new things or generate new ideas, but to take the existing expertise and democratize it by making it easier and cheaper for everyone,” Qureshi remarked. “I also think that we will see the biggest breakthroughs of AI mostly in team productivity, not in some game-changing products that integrate AI. You can have some interesting AI products only if you train your model on unique data, or get the model to do what no one has done before.”
Cheng pointed out that current AI models are not capable of solving complex problems. “For example, they can do basic things for security audits. Most of the current AI use cases are built around the application layer to leverage user experience in games, tools or productivity, but a lot of edge cases are still really hard to handle.”
Another problem with AI is that it can sometimes “hallucinate,” as Roman noted. “In some cases, this hallucination can have less impact, as in the case of fraud detection, but for some, such as autopilot cars with AI detection, it can have a much more significant impact,” he said.
Speaking of prominent startups using AI, Cheng highlighted the HIM project similar to the movie Her — where one can create and chat with a virtual boyfriend or girlfriend. HIM uses AI to allow each user to develop the boyfriend and the tone of the conversation they like, train virtual boyfriends to have and express emotions, and increase the pace of content generation.
“There are more interesting cases of AI blending with blockchain on the horizon. We’re excited to see what will grow from today’s ideas and entrepreneurs,” he added.
Disclaimer. Cointelegraph does not endorse any content or product on this page. While we aim at providing you with all important information that we could obtain in this sponsored article, readers should do their own research before taking any actions related to the company and carry full responsibility for their decisions, nor can this article be considered as investment advice.