OpenAI has already introduced that it’s going to simplify its ChatGPT product roadmap, which has grow to be horribly splintered these days by too many obscure product names for the common client to care about. It’s ditching all its newest mannequin names, comparable to o1 and o3, to not point out o3-high, to get behind a easy ChatGPT-followed-by-version-number branding type.
I feel it’s time Google {followed} swimsuit with Gemini, and if you wish to know why, simply have a look at this screenshot I took from the Gemini app on my telephone right now:
Speak about too many choices… (Picture credit score: Apple/Google)
It’s arduous to have a look at that screenshot and assume that anyone at Google deliberate it to look that manner; it’s simply the way it’s advanced, with little thought given to the one that has to make use of it.
As a part of the Gemini Superior subscription you get entry to all of Google’s completely different language fashions, nevertheless having to decide on between seven completely different fashions with very comparable names is verging on the ridiculous. To not point out that I feel ‘Google Gemini 2.0 Flash Thinking Experimental with apps’ might effectively win an award for the longest product title in historical past. What does that imply, anyway?
As a ChatGPT Plus subscriber, you additionally at present get entry to an equally baffling variety of completely different LLMs. Some, named issues like o1, are deep reasoning fashions that take some time to essentially mull over the questions you ask them, whereas others are smaller and extra light-weight, so that you get faster solutions however much less depth. It’s virtually like Google and OpenAI are competing with one another in how complicated they will make their AI instruments.
An excessive amount of selection
I get it; if you’re concerned in tutorial analysis or want AI that can assist you write code, you in all probability want entry to a wide range of completely different LLMs for various duties. But when all you wish to do is get AI that can assist you together with your child’s math homework or plan your subsequent highway journey, accessing such a various collection of LLMs is overkill.
OpenAI’s proposed resolution is to quiet down behind the ChatGPT title {followed} by a model quantity, so ChatGPT-4, then ChatGPT-5, and so forth. The AI will then have a look at the complexity of what you’re asking and determine if it wants to make use of a deep reasoning mannequin or an easier LLM. To torture an analogy, most individuals don’t care how the sausage is made; they merely need a delicious-tasting hotdog.
Google ought to do the identical with its completely different variations of Gemini. For instance, the ‘2.0 Flash Thinking Experimental with apps’ mannequin has the facility to entry issues like YouTube movies, so it could reply questions on them or extract the important thing factors from them. Why not simply name it Gemini 2.0, after which, in the event you ask a query a couple of YouTube video, change to the related LLM with out having to get the person to decide on it first?
Take a look at Apple
What we wish is AI that simply works, with easy naming conventions. If you happen to have a look at the way in which Apple manufacturers its merchandise, it usually has as much as three variations of a product with a easy title indicating the variations. It has two variations of its MacBook – the MacBook Air and MacBook Professional – and its newest iPhone – iPhone 16 and iPhone 16 Professional – that’s good and easy.
We’re nonetheless proper at the beginning of the AI revolution, and issues are altering quickly, and the buyer expertise is commonly left as an afterthought. As OpenAI’s resolution to simplify its branding exhibits, that scenario is beginning to change. I feel we will count on the expertise of utilizing AI to get rather a lot slicker this 12 months, let’s simply hope that Google get the memo, too.
You may also like
