What’s in a name?
Name Game
As OpenAI releases yet another version of its GPT large language model with an extremely cumbersome name, CEO Sam Altman has finally admitted that it might be time for a new naming convention.
After announcing on X-formerly-Twitter the release of the new LLM known as GPT-4o Mini, which will reportedly be cheaper for businesses to use in making their own chatbots than its non-mini version, Altman issued a rare response to criticism.
“You guys need a naming scheme revamp so bad,” quipped tech Youtuber Concept Central in response to the letters-and-numbers salad of its product names, which have previously included clunkers like “GPT-3.5 Turbo”
In perfect internetspeak, the OpenAI CEO responded “lol yes we do.”
Any Other Name
While there’s nothing explicitly wrong with the way OpenAI names its products, the company has branched out with the nomenclature for its other projects.
From its Sora video generator, which means “sky” or “void” in Japanese, to the DALL-E image generator that’s reportedly in homage to both the Pixar film “WALL-E” and Salvador Dali, OpenAI has clearly put thought into the names of its products.
Even the rumors we’ve heard for its mysterious upper-level AI projects, from Q* (pronounced “Q-star”) to the newly-reported “Strawberry” code name for the endeavor that’s said to be bent on imbuing LLMs with reasoning capabilities, there’s obviously no shortage of cool names being promoted during brainstorming sessions at the world’s leading AI firm.
By contrast, GPT isn’t even bespoke — it literally just means “Generative Pre-trained Transformer,” which is technically the classification for every LLM of that type.
We’ve reached out to OpenAI to ask if it has any renaming plans, but given how wildly popular ChatGPT and its undergirding LLMs have become, it’s kind of hard to see it happening anytime soon.
More on OpenAI: Most Users Think ChatGPT Is Conscious, Survey Finds
Source link
lol
