It’s now a year since OpenAI first announced the concept of the GPT Store and around nine months since the store officially opened. In that time, an estimated three million GPTs were launched and shared on OpenAI’s storefront, covering a range of services and expertise. Yet, if we are frank about it, you really don’t see much coverage of the GPT revolution that we were promised. All is not lost – far from it – but there are some questions to be asked about their relevance and, indeed, their future.
For those not aware, a GPT, at least in this sense, is a kind of customizable AI assistant, one that is supposed to be specialized in a particular field. If you want to talk to a chatbot about general topics, then you should use ChatGPT. If you want specialized knowledge, you would either access a GPT on the GPT Store or create one yourself by training it on specialist data.
From the outset, the concept of the GPT store was hugely exciting. It created the promise of a revolution in, for want of a better term, microservices. You could theoretically create a GPT for anything: As a companion for translating the works of JRR Tolkien; for strategy advice for playing online craps; a tool to help you market your restaurant; a travel companion to help you plan a trip to Asia; basically, whatever you could come up with.
To an extent, OpenAI delivered on that promise. You can use specialist GPTs for every task imaginable. If, for example, you put the word “Tolkien” into the search bar, you come up with numerous Tolkien-themed GPTs, ranging from tutoring for learning Elvish to a creative assistant for Tolkien-esque fantasy writing. There are options for different languages, too.
A GPT For Every Service You Can Imagine
There is a lot to discover on the GPT Store, although it did encounter some problems. For a start, by the spring the store was filling up with spam, with TechCrunch reporting that GPTs that ostensibly offered, say, image generation in the style of Disney, would serve as “funnels to third-party services” with little relation to the original purpose of the GPT.
And then, there was the question of copyright. OpenAI must pay huge sums to companies to access copyrighted material, but the same restraints aren’t placed upon GPT creators. In short, it turns out that training your GPT on the works of Tolkien might be a copyright problem. A variety of artists, publishing firms, and media organizations decided to get litigious.
Yet, arguably, the biggest issue on the GPT Store is discoverability. The UX of the platform is fairly basic, and it’s fine if you are looking for something specific, but the setup makes it difficult to discover smaller ‘creators’ without knowing what they are.
Some have argued for a TikTok-style feed for exploring GPTs. Whether this is the right approach or not is not clear, but it does feel like there should be an overhaul.
Lost In The Mix Of Constant AI Announcements
In the end, however, we might also argue that the GPT Store has been lost in the mix. Every week, there seems to be a new release of an LLM or AI service. It’s not just OpenAI, of course: Google Gemini, Meta Llama, Anthropic’s Claude, and even xAI’s Grok have been constantly launching bigger and better models and experimenting in different areas. It’s difficult for the tech media to keep up with.
However, the underlying ambition of the GPT Store should not be dismissed. There are issues to iron out, sure, and it’s arguably the case that it’s not going to be the vaunted “app store for AI services,” at least not yet. Yet, there is enough to be excited about, and perhaps we shouldn’t rule out the GPT Store as being revolutionary just yet.