Carl T. BergstromFeb 13, 2023

Meta. OpenAI. Google.

Your AI chatbot is not *hallucinating*.

It's bullshitting.

It's bullshitting, because that's what you designed it to do. You designed it to generate seemingly authoritative text "with a blatant disregard for truth and logical coherence," i.e., to bullshit.

Show threadGlen AkinsFeb 13, 2023
@ct_bergstrom perfect for writing hollow but good sounding corporate pr flack pieces. Or the trader joe's fearless flyer product descriptions.
Show threadDezene Huber 🌻Feb 13, 2023
@bikerglen @ct_bergstrom Our university is going through a "strategic planning process" so I asked it to write up a strategic plan and guided it along with a few of our institution's general parameters. The output text read like I expect the eventual true product will read... and like every other university strategic plan that I've read before with a few institution-specific spice words. (We could save a lot of money on consultant fees this way.)
Show threadTom Bellin 
@dezene @bikerglen @ct_bergstrom Maybe the insight is that you don't need the plan document at all.
Feb 13, 2023 at 11:51pmWeb
Show threadDezene Huber 🌻Feb 13, 2023
@tob @bikerglen @ct_bergstrom The only benefit that I can see from these exercises are that they keep senior administrators and their small empires busy, and not inventing new forms for faculty and staff to fill out.
Show threadTom Bellin Feb 14, 2023
@dezene @bikerglen @ct_bergstrom You definitely don't want them automating their processes using AI.
Show threadMartin Vermeer (not Jesus)Feb 14, 2023
@dezene @tob @bikerglen @ct_bergstrom If anything good comes out of this, it may be a better understanding of the mechanics of, also human, BS production. And for some folks, developing a better eye for spotting it.
Show threadJim ProctorFeb 14, 2023
@martinvermeer @dezene @tob @bikerglen @ct_bergstrom Is there an algorithm for measuring the extent to which a human text, say a campus strategic plan, *deviates* from a series of AI attempts? Then we might grade ourselves on originality.
Show threadTom Bellin Feb 14, 2023

@jimproctor @martinvermeer @dezene @bikerglen @ct_bergstrom IMO, this is the (awkward) way forward for these LLM-based attempts at "AI". Take the LLM generated content, run it through some sort of meta-analysis, and then present that to the user.

I mean, in a way they're already doing that to suppress the nazi stuff.

In your case, this hypothetical tool would be capable of doing a meta-analysis of your document and the LLM-generated document on the same subject.

Show threadTom Bellin Feb 14, 2023

@jimproctor @martinvermeer @dezene @bikerglen @ct_bergstrom Of course, you can't ask the LLM to do the meta-analysis. So we're relying on an as-yet non-existent technology to make the LLM useful in a practical way.

At which point, why are we using the LLM?

Show threadDezene Huber 🌻Feb 14, 2023

@tob @jimproctor @martinvermeer @bikerglen @ct_bergstrom

Really all I want is an AI that could retrieve, fill, and send forms to the next approval stage with a few simple typed or spoken commands.

Show threadgrob (teeth era) 🇺🇦🏳️‍🌈🏳️‍⚧️Feb 14, 2023
@tob @dezene @bikerglen @ct_bergstrom one can dream