While the recent advances in generative AI have led to the creation of some amazing tools, many of them lack the refinement to be truly transformative tools in professional industries. Sure, ChatGPT can produce human-like prose on any subject, but it is human-like, not quite good enough to replicate real writers, while it’s also prone to errors and mistakes that require expert oversight to avoid.
Which is similar in most of its use cases, ChatGPT and other generative AI tools can create good enough content, in some contexts. But to be truly valuable, they still require knowledge and refinement, so they’re handy as complementary tools, but not as valuable as some have suggested.
But the latest announcement from OpenAI could change that.
Today, at its first-ever developer conference, OpenAI announced a new slate of custom tools, built on its GPT back-end, that will be more customized to specific tasks and industries.
As reported by The Verge:
“In the coming weeks, these AI agents, which OpenAI is calling GPTs, will be accessible through the GPT Store. Details about how the store will look and work are scarce for now, though OpenAI is promising to eventually pay creators an unspecified amount based on how much their GPTs are used. GPTs will be available to paying ChatGPT Plus subscribers and OpenAI enterprise customers, who can make internal-only GPTs for their employees.”
Why is that a big shift?
Because right now, ChatGPT, and other generative AI tools, have a knowledge base that’s built on huge combined datasets, so that the system has more info to sort through for semantic matching, and understanding relevant meaning. But what if, for example, you could transfer that same processing capacity to more confined, more specific datasets, like a “Harry Potter GPT”, which is trained only on Harry Potter books, and could then create text in that same specific writing style, and within the parameters of that world?
What if you had a sports writing ChatGPT trained only on examples from sports publications and podcasts? A Renaissance painting creator trained only specific artists and works?
The more customized the system is, the more refined the outcomes will be, relative to the input. And an extension of that, then, would be custom AI bots, built using OpenAI’s code, and trained on only your own, internal documentation and insights.
That type of AI could be infinitely more valuable, within given parameters and scope, and that’s where OpenAI is heading with this new shift.
Meta is also exploring the same, with its AI Studio project, which will eventually enable people to create their own bots and automated engagement systems, based on their own communication style and voice.
Which could facilitate a range of more specific use cases, with answers better aligned to specific needs and requirements within a certain sector, topic, or industry.
How functional each system will be remains to be seen, and as noted, there will be base-level requirements that not all industries and businesses will be able to meet. But it could be the start of the next big generative AI shift, which will lead to more functional, valuable generative AI tool, at least in a professional sense.
Which could then open up new avenues for customer assistance chatbots, messaging tools, promotions, etc.
There’s a range of possibilities, and it’ll be interesting to see what becomes available in the new GPT Store.