A humble blog post in November from tech startup OpenAI introduced the world to ChatGPT.
Since then, the artificially intelligent conversation program that can sound convincingly human has taken the world by storm, appearing in hundreds of articles, threatening to upend Google’s search dominance, and even writing a comedy bit performed by Billy Crudup on Jimmy Kimmel’s late-night show. ChatGPT already has 100 million monthly active users, making it possibly the fastest-growing app ever, according to UBS.
Not surprisingly, chatbots, and the underlying technology area known as generative AI, are also stoking interest among investors and entrepreneurs. In addition to text-based chat, startups are looking at using generative AI for making original images, musical compositions, videos, and more.
“This is not in the research lab anymore,” said Rana el Kaliouby, deputy chief executive of Swedish AI company Smart Eye (and cofounder of Boston AI startup Affectiva). “I think every company is trying to figure out how to adopt these technologies now.”
Boston’s numerous university programs focused on AI have helped drive a new wave of startups working on generative apps that have already attracted financial backing. And many more are likely on the way, said Brian Smith, a Boston College computer scientist and associate dean for research.
ChatGPT “hasn’t been out that long and people are already going, ‘How do we make money off this,’” Smith said. “There’s definitely lots and lots of people looking at it now.”
Several local companies got started before the hype. Common Sense Machines in Cambridge emerged from MIT in 2020, founded by cognitive scientist Josh Tenenbaum and two of his former grad students, Max Kleiman-Weiner and Tejas Kulkarni. The startup, which employs 10 people, raised $5 million in seed funding with a focus on using AI to generate 3-D environments.
Based on Tenenbaum’s research into how babies’ brains develop, the startup is trying to enhance the creativity of its app by adding a bit of, well, common sense. Human babies form an understanding of the world by developing abstract models. Common Sense Machines is trying to add similar models to its 3D world builder.
“Think of it as an AI copilot for actually creating and coding a 3D environment,” Kulkarni said.
At least for now, the startup is focused on helping video game designers and others who don’t need the pinpoint precision of architects or mechanical engineers.
Novus Writer in Boston, a text-focused generative AI startup, raised several hundred thousand dollars in seed funding last year and employs about 15 people. It is focused on using AI to write marketing copy for websites, ads, and other pitches. Aimed at business users, the company has about 100 customers so far, Egehan Asad, cofounder and chief executive, said.
The founders include grad students from MIT and Northeastern who knew each other from their undergrad days at Koç University in Istanbul. At first, Novus was developing apps for marketing media and got early backing from the MIT Sandbox program. The company pivoted to using generative AI a year ago.
Novus is using multiple AI models in an effort to avoid some of the pitfalls that have plagued ChatGPT, such as producing convincing but factually incorrect answers. In addition to fact-checking, the company also wants to prevent the app from plagiarizing content from the web or from its own previous answers.
“People who are new to AI are expecting something magical,” Asad said. “But in the end if you are creating an output on a specific topic, it is very possible you are getting the very same result as some other person.”
ChatGPT and other generative AI programs have attracted as much attention for their mistakes as for their abilities. And that may be just the opening for Novus and other startups to build a business.