Could generative AI be coming to Twitter as well?

While repeatedly warning people about the dangers of AI development, Twitter chief Elon Musk is apparently exploring AI options for his new company, with the billionaire buying up thousands of GPUs as part of a larger initiative designed to develop his own AI tools to compete with OpenAI.

As reported by Insider:

“Musk's AI project at Twitter involves a large language model, according to one of the people familiar. An LLM is essentially AI that is trained on large sets of data so it can eventually create new, seemingly sophisticated content and text on its own. Twitter has massive amounts of data that could train an LLM.”

Indeed, OpenAI initially used Twitter data to train its GPT models, before Twitter cut off its access late last year. And now, Musk and Co. are exploring their own, competing project, which would use Twitter’s massive troves of data to fuel its own AI models, with the intention of staying up to speed on the latest trending tech development.

Which is not a huge surprise. Last month, reports emerged that Musk was looking to recruit AI developers to build his own, ‘anti-woke’ AI model, amid rising concerns about speech and topical restrictions in ChatGPT and other tools.

This is just one example of how Musk sees AI tools as potential agenda-shifting models, which is a key reason why he’s now looking to build his own AI tools to compete.

That ties back to his broader push on free speech, and providing a platform for all viewpoints, regardless of your personal inclination towards one side of politics or another. Musk has repeatedly claimed that this is his key motivator for buying Twitter, in an effort to provide a more open platform for all, and along this line, his development of a new, separate AI model makes sense.

Musk also has beef with OpenAI, a company that he was an early investor in, before it launched its consumer tools. Musk donated $100 million to OpenAI in its early stages, as a non-profit, and he’s repeatedly vented his frustrations about the company becoming a for-profit business ever since.

That’s likely another motivator in Musk’s AI push - though how exactly AI might fit into Twitter remains to be seen.

Could you imagine an AI model built entirely on tweets? The angst and toxicity alone would be a sight to behold.

As noted, Musk has also warned of the risks of AI development, recently signing an open letter calling for a 6-month pause on AI advancement. Though I’d hazard a guess that Musk’s ‘concern’ was more driven by spite, and trying to stop OpenAI from making money based on his initial investment (from which he won’t see a dime).

Either way, we could find out soon – Musk has reportedly hired two former DeepMind engineers for the project, while also spending tens of millions on the new GPUs required to build the system.  

Will that lead to more generative AI tweets? More customer service tools in Twitter DMs? Visual prompts that respond to tweet prompts?

There’s a range of potential options, and given the scope of investment, it could be on the way in the very near future.