Google reminds everyone it too can launch a ChatGPT-like chatbot … waiting list Google is offering Bard – its chat-driven rival to ChatGPT – to netizens in the US and UK who ask nicely. Bard is derived from the web advertising giant's large language model LaMDA, and was built to compete with OpenAI's GPT series – the brains behind the chatbot interface for Microsoft's Bing search engine, 365 suite, and …
That said, I'm going to take issue with this:
As a non-intelligent information regurgitation engine, it doesn't really know the answer to a question, nor understand the actual problem – it just draws from what it was trained onYou're implicitly endorsing the position of John Searle and his Chinese Room.These programmes clearly have functional understanding good enough to give answers, for many cases. Yes, they were trained, but so were we. Yes, they give wrong answers, but so do we; we have whole genres of entertainment built around our inability to give correct answers to tricky questions. Yes, they bullshit and don't tell us they're fabulating, but so do plenty of us, and we frequently put those people in charge.
We don't understand our understanding or understand their internal models well enough to say whether they understand to the degree and in a similar manner to how we do. This is why we had the Turing Test: the answer to whether they understand should be based on the functional appearance, not the mechanism used to achieve that and our prejudices about it.
Thus far they, seem incapable of coherently reasoning - and given our understanding of how they are built, it seems unlikely this class of models will ever do that well. (But never say never.) And as I say, there's more to intelligence and being a sentient being than understanding facts and questions. But maybe our own understand is no more than a glorified auto-complete, too. And maybe saying they don't understand is to display our own pro-human bias.
You're missing the one major difference between us and GPT-alikes – we would rapidly get fed up with people asking stupid questions and ordering us to write essays or limericks or fake research papers, and we'd bugger off and do something that we wanted to do.
In 10 years there will be no need for a license to any Adobe product but this one. Real digital art pros will have moved on to more complex, non-Adobe 3D applications while pro advertisers will use point and click menus in Adobe's generative tool.
The "Do Not Train" B.S. that Adobe is spewing about is so Adobe can lock-in customers. It's inevitable that Adobe will soon have a consent form that will allow Adobe to use content created by its customers for its own commercial purposes. Of course that content will have a "Do Not Train" tag on it so that the original customer cannot themselves use their own content in the Adobe derivative. Worst case for Adobe is that they've eliminated illustrations jobs in favor of pushing people to their generative tool... it's a Win-Win for Adobe.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
