[flagged]
> Please don't post shallow dismissals, especially of other people's work.
GPT-5 Thinking in ChatGPT (a.k.a. Research Goblin) is good at search
31–40 of 268 posts
Re: GPT-5 Thinking in ChatGPT (a.k.a. Research Goblin) is good at search
#32Earlier quoted context omitted.
Spending thousands of words to essentially say "ChatGPT's search feature works pretty well now" with mundane examples like finding UK cake pop availability or identifying buildings from train windows. This has been done before by less capable models - it's just a rehash. Should we expect newer models getting worse? The breathless "Research Goblin" framing and detailed play-by-play of basic web searches feels like pad…
The mundane examples were the point . I'm not picking things to show it in the best possible light, I picked a representative sample of the ways I've been using it. I called out the terrible scatter plot of the latitude/longitude points because it helped show that this thing has its own flaws. I know so many people who are convinced that ChatGPT's search feature is entirely useless. This post is mainly for them.
[flagged]
Re: GPT-5 Thinking in ChatGPT (a.k.a. Research Goblin) is good at search
#33Off topic, but I wonder why the author is using _both_ Substack and his old website [1]. Is this a new trend?
Re: GPT-5 Thinking in ChatGPT (a.k.a. Research Goblin) is good at search
#34Yeah, the % of the time I need to dip into deep research with GPT5 is much lower than GPT4 for sure. It even beats Gemini's web grounding which is impressive, I think most of the lift here is how smart/targeted its queries and follow-ups are.
Re: GPT-5 Thinking in ChatGPT (a.k.a. Research Goblin) is good at search
#35Re: GPT-5 Thinking in ChatGPT (a.k.a. Research Goblin) is good at search
#36[flagged]
HN is very cult-of-personality based. People see SimonW they upvote without reading, while at the same time a much better article could be posted on the same topic and get zero traction. Not trying to single Simon out here, I generally find his posts good, just a statement of the herdthink and cognitive laziness of this community (and humans in general, to be fair).
Re: GPT-5 Thinking in ChatGPT (a.k.a. Research Goblin) is good at search
#37Earlier quoted context omitted.
Spending thousands of words to essentially say "ChatGPT's search feature works pretty well now" with mundane examples like finding UK cake pop availability or identifying buildings from train windows. This has been done before by less capable models - it's just a rehash. Should we expect newer models getting worse? The breathless "Research Goblin" framing and detailed play-by-play of basic web searches feels like pad…
Simon says “what I used to Google I now try AI thinking models” I didn’t feel that he was framing it as _revolutionary_ it felt more evolutionary . Simon, for every person miffed about your writing, there is another person like me today who said “ok, I guess I should sign up for Simon’s newsletter.” Keep it up. It’s easy to be a hater on da internet. 42lux, if you have better articles on AI progress do please link th…
[dead]
Re: GPT-5 Thinking in ChatGPT (a.k.a. Research Goblin) is good at search
#38Earlier quoted context omitted.
The mundane examples were the point . I'm not picking things to show it in the best possible light, I picked a representative sample of the ways I've been using it. I called out the terrible scatter plot of the latitude/longitude points because it helped show that this thing has its own flaws. I know so many people who are convinced that ChatGPT's search feature is entirely useless. This post is mainly for them.
[flagged]
The thing about models getting incrementally better is that occasionally they cross a milestone where something that didn't work before starts being useful.
Those are the kinds of things I look out for and try to write about.
Re: GPT-5 Thinking in ChatGPT (a.k.a. Research Goblin) is good at search
#39Off topic, but I wonder why the author is using _both_ Substack and his old website [1]. Is this a new trend? [1] https://simonwillison.net/2025/Sep/6/research-goblin/
I use Substack as a free email delivery service - it's just content copied over from my blog: https://simonwillison.net/2023/Apr/4/substack-observable/
Re: GPT-5 Thinking in ChatGPT (a.k.a. Research Goblin) is good at search
#40Nice writeup. This may nudge me to start using chatbots more for this type of queries. I usually use Perplexity or Kagi Assistant instead. Simon, what's your opinion on doing the same with other frontier systems (like Claude?), or is there something specific to ChatGPT+GPT5? I also like the name, nicely encodes some peculiarities of tech. Perhaps we should call AI agents "Goblins" instead.
I've been much more impressed by GPT-5 than the other systems I've tried so far - though I did get a surprisingly good experience from the new Google AI mode (notably different from AI overviews): https://simonwillison.net/2025/Sep/7/ai-mode/