#ai is fail

[generatingfail]

GenAI-free search engine: noai.duckduckgo.com

I just learned (thank you, Mastodon) that the DuckDuckGo search engine has a subdomain https://noai.duckduckgo.com/ that leaves out all the GenAI nonsense! No muss, no fuss, no bullshit machines!

But, you ask, how do I ensure I'm always using it? Fair question. If I tried to do this manually, I would fail hard.

How to auto-forward the regular DDG domain to this one depends somewhat on your browser. Here's how I did it in Firefox desktop:

Add the Redirector add-on to your Firefox. Go into its preferences ("Edit Redirects") and add a rule that changes the Include Pattern: to https://duckduckgo.com/* and the Redirect to: pattern to https://noai.duckduckgo.com/search?q=$1

For me, this is Just Working so far, including from the browser's own search bar.

I'm extra-happy to be signaling to DDG that the bullshit machines can take a flying leap.

[cuphead030]

Turn this on! I don’t want ur guy’s art stolen by ai!!! It’s in visibility in settings!

I love u guys 🥺🫶🏻🫶🏻

[generatingfail]

Saying the quiet part aloud, I see.

[generatingfail]

Report on AI benefits contains AI slop

From the even-a-chatbot-can't-make-this-shit-up files:

Major consulting and analysis firm KPMG put out a glowing report on how generative AI has transformed society.

Turns out a bunch of it was lying made-up AI slop.

KPMG has withdrawn the report. (If anybody has a copy, though, I'd love one -- DM me for where to email it.)

Source (paywalled, excerpt in this Mastodon post)

[generatingfail]

The Emperor's New Image

Please describe the image above this paragraph in this post.

Go on, describe it.

If you just said "there's no image there, nitwit!" you're not a chatbot.

Because chatbots, it turns out, will happily and verbosely describe images that weren't given to them.

The researchers who discovered this called it "mirage."

This will probably be fixed quickly -- it's not hard to get a computer to check for an image (unless you're vibe coding) -- but for now, it's freakin' hilarious.

[generatingfail]

Responding to Controversy, Seattle Worldcon Defends Using ChatGPT to Vet Program Participants

Seattle Worldcon 2025 Chair Kathy Bond today issued a public statement attempting to defend the use of ChatGPT as part of the screening proc

So, let me get this straight. The leadership at Seattle's Worldcon, a SFF fan convention that is also hugely important to working creators:

used a chatbot service widely excoriated by working creators because it plagiarizes them and openly competes with their labor...

... to vet potential panelists, when its parent corporation is defending defamation lawsuits because it allegedly emits damaging lies about people...

... not to mention the service is demonstrably biased, in an area of publishing that absolutely doesn't need any more bias...

... without telling anyone.

Heckuva job, Kathy. Let's see if you manage to do as much damage to the Hugos as the Puppies did. Looks like it'll be a close race.

Additional commentary at Pivot to AI.

[generatingfail]

LLMs in SFF are fail -- tell SFWA this!

The Science Fiction Writers Association has published a survey about the ethics of LLM use. Anyone can take it, but I suspect creators' views will be given additional weight.

It's not the typical push poll one often sees about AI, in which "I don't use it and don't think anyone else should" doesn't even appear as an option. It lets you say that you don't! It lets you say that you think LLMs should be banned! Its one significant content problem is that it doesn't explicitly acknowledge the existence of fan creators, but there are free-text fields where that gap can be raised and questioned.

(It also asks for your name and email address. If you refuse to take it on that basis, I can't quarrel with your decision.)

Go take it! I did.

This post published 4 January 2026.