User talk:Jimbo Wales

Thoughts on AI chatbots?

Hi Jimmy, I was wondering if you use AI chatbots. I jump between multiple AI chatbots every day. If you use AI chatbots, which one do you prefer or do you use multiple? I've used ChatGPT a lot in the past, but lately, I've been getting frustrated with it, not because of the Pentagon deal, but I find other AI chatbots to have better responses especially with current information. What do you think? How do you see the AI race going in the years and decades coming up? Interstellarity (talk) 00:15, 8 April 2026 (UTC)

I use several of the major ones daily, and I'm also very keenly exploring open source local ai. ChatGPT, Claude, Gemini are things I test/use every day. The "best" changes over time as they release new models and features.

I'm also eagerly exploring OpenClaw (don't worry, I'm not giving it access to my email or my computer!) for the opportunity to do more persistent "agentic" things.

My feeling today, after spending a fair amount of time with ChatGPT trying to solve a computer problem, is that chatting with a chatbot is actually to some extent misleading people about the potential here and what they are best at. I've used AI chatbots to work with me on some mid-sized fun hobby coding projects, and that process is both amazing and frustrating. But I'm just starting to work using OpenClaw to do 'agentic' coding where you have the agent create tests, code to the tests, etc. It's... pretty amazing.

One example (which I am not yet working on but see a path to quite easily) that will be of interest to Wikipedians: a frequent question around NPOV is trying to sum up in a straightforward way "What do the sources say about this?" Think of a relatively straightforward question like "What pronouns do the sources use to refer to this person? What is the date on the sources?" To create a list of every source in an article in a table which says what pronouns are used, what the date of the source is, what the source is (type of source) is a lot of tedious manual labor but which can be done in an objective way for the most part. And it would be useful to help others get oriented to the situation, as a useful artifact to help the discussion move forward productively. For a human to do it - it's a lot of work. For a bot working in a step by step process, checking its work multiple times, it'll cost something but it will be pretty good and can be checked by a human and (to some extent) by a more traditional script. (For that last, what I mean is, if the bot gives an exact quote from a source it could possibly be hallucinating, but because the claim that the exact quote appears in the source can be checked by a non-AI script with some certainty.)

The result would not necessarily be unassailable. Errors, while rare, could still happen, just the same as if a human did it. So there might be some refinement and discussion about what the results mean. But it strikes me as a potentially useful thing that anyone could do to help move a discussion forward. Does this help with every discussion? No. But in some cases it could be useful.Jimbo Wales (talk) 21:17, 8 April 2026 (UTC)

A breath of fresh air after my eyes glazed over in the face of all the AI doomer spoutation at the Village Pump. Carlstak (talk) 04:17, 9 April 2026 (UTC)

I am one that keeps up with the AI news all the time and always keen to see the latest features coming to new and existing models. As an AI enthusiast, I am always looking for ways I can incorporate AI in my day to day life and helping my family implement these features. I've been following what Google has been doing since they do a lot of amazing stuff with AI. They really transformed search into an answer engine similar to Perplexity with AI Mode. I enjoy reading up on all of these to see where the industry ends up heading. Interstellarity (talk) 01:04, 11 April 2026 (UTC)

@Carlstakand @Interstellarity do you know somewhere (a wikiproject perhaps) where there might be a grouping of people interested in and/or actively exploring some simple tool building with AI that might be useful in English Wikipedia in particular? I'd like a better way to share things I'm poking around with, and also hear from others as to what they are trying out.

Regarding the AI doomerism I think it's important to acknowledge that very bad and annoying uses of AI are happening in real time today. But rather than hand waving and talking, I think those of us who think it can be a useful tool should get together and roll up our sleeves to start building things that people will like. And I assume that's already happening in a disjointed way but it'd be good to collaborate. Jimbo Wales (talk) 14:09, 14 April 2026 (UTC)

The very aptly named WP:WikiProject AI Tools is probably the place to go. CMD (talk) 14:29, 14 April 2026 (UTC)

I'm also curious about what devices you own. What kind of a phone do you have? Are you an iPhone or an Android user? What about a computer? What kind of a computer do you have? Do you have a Windows, Mac, or a Chromebook, maybe you are into Linux if that's your thing? I personally own an iPhone 16 while I own a 2019 Dell XPS laptop running Windows 11 although at some point in the future, once my computer reaches the end of its life, I might consider getting a Mac since I already own an iPhone. What do you think? Interstellarity (talk) 00:19, 15 April 2026 (UTC)

Please forgive me, but this is kinda funny. In response to Jimbo, the page CMD (does he live in a Wikimedia server?) pointed to looks like the place, based on all the usernames I see there. Carlstak (talk) 01:07, 15 April 2026 (UTC)

My daily driver is a Macbook M4 Max with 128GB of RAM purchased at vast expense when it came out. Its unified memory design has allowed me to run larger models albeit somewhat slowly. I also have a couple of cheap mini pcs running kde/linux for example one at my treadmill where I exercise and watch various things. I also have a NAS running proxmox with a number of virtual machines.

And finally I just bought a AMD Strix Halo mini pc (Corsair) with 128GB of ram which I'll be setting up to run OpenClaw and (probably) Gemma 4 as a test of how good agentic programming restricted strictly to local AI is getting. I'll start using it by attempting to have it code up some silly little website ideas that I have - it's as much to get hands on with that as a real need or desire to do it in and of itself.

In terms of phones, I'm an android man since forever. I tend to buy cheap phones with huge batteries because it allows me to be a dad hero and share charge with family members in the evenings with their puny little iphone batteries (ha!). Jimbo Wales (talk) 10:05, 15 April 2026 (UTC)

@Jimbo Wales can i advise you to be careful with sharing too much of that kind of information ? It makes it easier to send targeted hacking attempts your way. —TheDJ (talk • contribs) 11:04, 15 April 2026 (UTC)

Well, it's pretty standard hardware, but ok. :-) Jimbo Wales (talk) 12:38, 15 April 2026 (UTC)

@CMD this is perfect, thank you! Jimbo Wales (talk) 10:06, 15 April 2026 (UTC)

yeah, ive tried gemini search for Linux troubleshooting with a moderate amount of success, and it franky probably failed to fix my issue because my specific distro paired with SSSD is kinda unusual. plus RHEL support forums being paywalled now. the thing to remember, is that AI isn't coming up with anything new per se, rather, it's predicting or working as a really good search engine most of the time. I do think many people don't really understand that. -- Aunva6^{talk - contribs} 03:45, 11 April 2026 (UTC)

Welcome!

Hi @Jimbo Wales! Wishing you a warm welcome to Wikipedia:WikiProject AI Tools – we are more than happy to see your expertise here! The project is still relatively recent, and as you can see we're still organizing some things, but source verification is definitely something which AI could help with. Human oversight is of course necessary, but given the very scale we're dealing with (including due to, you guessed it, AI spam), having tools to flag possible issues to check first would save us a great deal of time. Chaotic Enby (talk · contribs) 11:31, 15 April 2026 (UTC)

PS: As you might have guessed, I co-founded both WikiProject AI Cleanup and WikiProject AI Tools. The two go hand-in-hand surprisingly often, and you shouldn't be surprised to see the same folks working against AI slop and figuring out how to harness AI for good against it! Chaotic Enby (talk · contribs) 11:34, 15 April 2026 (UTC)

Wikipedia's general take on popular culture (outside Fandom)

Hello, Jim. I've been listening to your recent interviews, like ones on podcast, and you seemingly have deep interests in politics and social issues. Enough about that, I've seldom or barely heard you discuss pop culture, famous figures of pop culture, and Wikipedia's handling of such unless it's related to Fandom... am I right? I think seeking one of your interviews about pop culture, especially on Google or WP:The Wikipedia Library, would be a slog, and I'm hesitant to ask for one at WP:REFDESK.

I know that "pop culture" is a broad term, but I think Wikipedia's handling of pop culture would be nice to hear in your interviews and more inviting, especially to potential listeners and editors. Generally speaking, why more on politics and less on pop culture? George Ho (talk) 09:25, 16 April 2026 (UTC)

I answer the questions I get asked is probably the main reason? You won't find me opining much on chemistry or ancient Egypt either, right? I try to only talk about things that I know about, such as for example Internet public policy.

But having said all that, I have no objection to answering questions about pop culture! Not as some kind of expert on it, but I can talk about what culture I consume, how Wikipedia handles such issues, etc.

I do sometimes talk about a few cases. For example, when Michael Jackson died we were perhaps surprisingly slow to report it, it was reverted and kept off the page for a few hours as I recall - because there wasn't confirmation in reliable sources. I think that's pretty good!Jimbo Wales (talk) 12:41, 16 April 2026 (UTC)

Okay, the death of Michael Jackson was kinda seventeen years ago, wasn't it? Anything else related to pop culture in span of the last seventeen years that has crossed your mind?

Lemme guess... Gamergate, right? We all know how that turned out... Wiki-chaos, off-wiki talks, ArbCom case (which I read not exactly throughly but looked through long, long lists of stuff and all), etc. Not to mention the media coverage on Wikipedia's take on Gamergate... Still wanna go there, huh?

Besides Gamergate, anything else pop culture-related within the last seventeen years that has crossed your mind?

...Or, you'd still like to discuss Wikipedia's 2000s coverage of pop culture? Well, we've known already how Wikipedia itself had developed in the 2000s, right? George Ho (talk) 15:04, 16 April 2026 (UTC)

Crap... Almost forgot to ask about Wikipedia's coverage within its last seventeen years on popular culture, old (i.e. from more than seventeen years ago) and contemporary (i.e. within the last seventeen years). George Ho (talk) 15:08, 16 April 2026 (UTC)

I can't speak for Jimbo, but there are factors of pop culture that make it harder to.cover compared to other topics: reliable sourcing, and enduring coverage. The majority of sources that cover pop culture are of the clickbait variety, and rarely qualify as an RS for en.wiki purposes. Secondly most pop culture elements very much short term fads that come and go in days, failing the standard for notability and enduring coverage. Both make it hard to give serious encyclopedic discussion of these.

Mind you we have problems in areas like politics where there is far too much focus and detail coming from primary sourcing (newspapers and such), and editors write at a far too detailed level and not where encyclopedic summaries should be, so id not use politics as a metric. Masem (t) 12:45, 16 April 2026 (UTC)

To exemplify what you said, this reminds me of how hard covering the Cheers episode "The Gift of the Woodi" was when I started it... well, as a draft before an AFC volunteer approved it into the main namespace. Then I found out its prior AFD history, further showing how amateurish we were as volunteers when it comes to covering subtopics of pop culture. George Ho (talk) 14:44, 16 April 2026 (UTC)

After one point near the early days of WP we had an article for every known Pokémon at the time, but eventually we "matured" recognizing that wasn't well covered content for an encyclopedia, hence the WP:POKEMON test. That would eventually lead to notability and a means to determine, with a lot of gray space, when a topic was covered well enough to be a standalone topic. And through that much of the early pop cultural coverage was merged to more compact articles or deleted to reflect the coverage.

Now this is not saying we can't cover pop culture but we have to wait to make sure the element is enduring. Eg we have a standalone on the 6-7 meme as a recent example. Masem (t) 15:30, 16 April 2026 (UTC)

Should've been Wikipedia:Pokémon test... George Ho (talk) 17:22, 16 April 2026 (UTC)