I tried out the Deep Research feature for it today using my work account. It worked pretty well, not perfectly, but well enough. I chatted with it after it completed its report and it was able to hold the research and the report itself in context to answer my follow-up questions quite comprehensively.
When I got my new phone recently, I asked of it what is by a wide margin my most common voice task; setting a timer for something I was cooking.
It presented a UI suggesting it had understood the assignment, but utterly failed to actually set the timer.
It was at that point I reverted to Assistant and forgot it existed.
This feels par for the course though; a bunch of effort spent on a few “hard” tasks to make it seem impressive, but zero on maintaining existing functionality that normal people actually use on a regular basis.
Yup, my experience exactly. Couldn’t set timers or schedule meetings, and if it can’t do that it’s useless. So I switched back to assistant and never thought about it again.
Bad? From what perspective? I’ve been using it no problem. Sure, it can get delusional (what LLM doesn’t?). But it’s useful to get the general gist of something.
I don’t use llms myself, but in the past when I’ve heard of Gemini it was in relation to a mess up. It gave the impression that Gemini is even worse than all of the others.
That’s been my experience as well. It’s fucked up setting an alarm once, but I gave it some really unclear instructions in the middle of a conversation with my wife and it did understand something didn’t end up the way it was supposed to. ¯\_(ツ)_/¯
Isn’t Gemini really bad? I can’t recall ever hearing praise for it.
I’ve only used it as a search engine for finding public data sources and general business research.
The search result summaries and search results categorization style is helpful.
For this use case it works much better than ChatGPT and Le Chat.
I tried out the Deep Research feature for it today using my work account. It worked pretty well, not perfectly, but well enough. I chatted with it after it completed its report and it was able to hold the research and the report itself in context to answer my follow-up questions quite comprehensively.
When I got my new phone recently, I asked of it what is by a wide margin my most common voice task; setting a timer for something I was cooking.
It presented a UI suggesting it had understood the assignment, but utterly failed to actually set the timer.
It was at that point I reverted to Assistant and forgot it existed.
This feels par for the course though; a bunch of effort spent on a few “hard” tasks to make it seem impressive, but zero on maintaining existing functionality that normal people actually use on a regular basis.
Yup, my experience exactly. Couldn’t set timers or schedule meetings, and if it can’t do that it’s useless. So I switched back to assistant and never thought about it again.
I use it just for screen search/translation on my tablet and it’s excellent at that, much better than Bixby.
It’s like living in the future!
Depends on your point of view. It’s better than chatgpt, but still bad.
Bad? From what perspective? I’ve been using it no problem. Sure, it can get delusional (what LLM doesn’t?). But it’s useful to get the general gist of something.
I don’t use llms myself, but in the past when I’ve heard of Gemini it was in relation to a mess up. It gave the impression that Gemini is even worse than all of the others.
From privacy perspective
deleted by creator
People tend to ask personal questions when LLMs are readily available
That’s been my experience as well. It’s fucked up setting an alarm once, but I gave it some really unclear instructions in the middle of a conversation with my wife and it did understand something didn’t end up the way it was supposed to. ¯\_(ツ)_/¯