Google is for search, ChatGPT is to find.
My impression after my first encounters with ChatGPT.
For a query in a classic search engine, I know from many years of experience: I have to make an effort to choose suitable search terms, which I then combine by (unnaturally) stringing them together loosely - and in a way that I believe the machine can process well. Then I have a chance to rummage through the long list of results and come up with a result that brings me closer to a useful answer. That’s SEO on the user side, so to speak!
On the other hand, I text the ChatGPT with my questions or thoughts - just as they are buzzing around in my head. I then get an answer exactly as I understand it. Sometimes there are “aha” effects, sometimes there’s euphonious nonsense. (As things stand today, I should definitely make the effort to verify the answers). The machine optimises itself to the cognition of the user, therefore also to a pleasant interaction chat.
Examples of chat processes
- A short example of a specific software-related issue: ChatGPT - Install npm via PowerShell.
- Example of learning English with ChatGPT: ChatGPT - Learn Present Perfect.
- A longer interview-type example in two parts: ChatGPT - Fed's functions explained
My conclusion
Google is for searching, ChatGPT is for finding.
I am quite fascinated by this new technical possibility. I’m just as creeped out by the idea of the precision with which such a machine can create a psychogram from my natural language conversational behaviour. The Age of Surveillance Capitalism borders.
similar
A clever assessment of the current AI situation can be found in LNP