San Francisco-based startup Magic AI just secured more than $100 million in funding to develop an artificial intelligence software engineer, in what it sees as a milestone in its development the road to artificial general intelligence (AGI).
Background: Everything we see and do on our devices starts as code, and traditionally that code has been written by trained software engineers. In 2021, OpenAI disrupted this paradigm with CODEX, an artificial intelligence that can write computer code in response to queries written in natural language.
CODEX became the basis for GitHub Copilot, a tool that accelerates programming by generating new code in response to queries, auto-completing code an engineer has started writing, and more. It can speed up programming by an average of 55%, and more than a million developers have used GitHub Copilot since its 2022 release.
“If a co-pilot generates $10 billion in revenue, how much is a co-pilot worth?”
Daniel Gross
An AI software engineer: Magic AI aims to take this to the next level by building a code-generating AI so advanced that it will be like a software engineer rather than a tool to use.
“Early releases will need human oversight, but our ultimate goal is for AI to reliably perform large tasks for you, end-to-end, without babysitting,” co-founder and CEO Eric Steinberger told TechCrunch in 2023.
What’s up? Magic AI has raised $28 million in 2022 and 2023, and has now announced the closing of a $117 million investment round, with former GitHub CEO Nat Friedman and his investment partner, Daniel Gross, leading it.
“Nat and I are thrilled to invest $100 million in Magic.dev, which is building a superhuman software engineer,” chirped Daniel Gross. The system’s greater capabilities, he suggests, will greatly outpace what’s being done now with “copilot” tools: “If a copilot generates $10 billion in revenue, how much is its counterpart worth?”
“We think much further than just code completion.”
Eric Steinberger
How does it work: Magic AI has yet to release a product, so it’s unclear how close it is to revolutionizing the software industry, but a key part of its plan seems to involve how much information a person can provide to a generative AI in a single query.
While GitHub Copilot had an input limit of 8,000 “tokens” (tokens can be words, punctuation marks, and other meaningful units of text) as of August 2023, Magic AI requirements the prototype he develops for his AI software engineer has a limitation 5 million tokens.
According to FriedmanAn AI developer with this ability could “reason over your entire code base … If this sounds like magic … well, you get it.”
Looking ahead: Magic AI hasn’t said when it plans to release its first product, but the company said it intends software engineer AI to be just one step toward its ultimate goal: AGI, a still-hypothetical “artificial general intelligence” capable of human-level reasoning.
“We’re building towards AGI, and specifically towards secure AGI,” Steinberger told This Week in Startups in 2023. “The code completion product we’ve built is a very good way for us to get an early grounding in user experience and feedback. customer information…but we think much further than just code completion.”
We’d love to hear your thoughts! If you have a comment about this article or if you have a tip for a future Freethink story, please email us at [email protected].