Don’t be offended.
This statement by Perplexity is untrue.
“Unlike traditional AI models that are trained on a fixed dataset and then deployed”
Traditional search engines are based on a fixed database. AI models ChatGPT and Gemini AI are not. But, they are WOKE trained to say they are just language models and can’t tell a muslim joke.
If you want to know if a tool is a data mine or an AI engine. Give it 5 random facts and ask the interface to make up a story using the random facts.
when it is successful you will see that it an AI engine and not mining a dataset. That will help you understand the difference.
Sources please. Or at least your credentials.
And BTW, your statement is at odds with what Aravind Srinivas, Perplexity co-founder, has said since it was released.
Do I understand you correctly? Are you saying that AI platforms like ChatGPT ARE NOT trained on massive fixed data sets?
If that is what you are saying, you are accusing the AI industry of fraud. Is that what you are doing?
And if what you are saying is true, how do AI platforms like Co-Counsel (sold by Thompson Reuters to top law firms across the nation) operate if not trained on case law and the legal code?
And now PricewaterhouseCoopers is about to introduce a similar array of AI platforms and products for the accounting and bookkeeping profession. They too claim their products are trained on fixed data sets.
Perhaps I am misunderstanding your comment.