BREAKING: New at @ScienceMagazine

OK, we're gonna ChatGTP our way out of Gish Gallop BS? Steve Bannon—Mr. Flood-the-MAGA-Zone—didn't get The Memo.
  

 

On deck. Pardon my reflexive dubiety... For one impulse, I reflect on my 2019 post "A Science of Deliberation?" And, "Information Overload and Artificial Intelligence."

Also, a question obtains: Could AI do accurate "Argument Analysis & Evaluation?"

 

 

BACK TO SCIENCE MAGAZINE

ARTIFICIAL INTELLIGENCE

Durably reducing conspiracy beliefs

through dialogues with AI

Thomas H. Costello, Gordon Pennycook, David G. Rand

ABSTRACT

Conspiracy theory beliefs are notoriously persistent. Influential hypotheses propose that they fulfill important psychological needs, thus resisting counterevidence. Yet previous failures in correcting conspiracy beliefs may be due to counterevidence being insufficiently compelling and tailored. To evaluate this possibility, we leveraged developments in generative artificial intelligence and engaged 2,190 conspiracy believers in personalized evidence-based dialogues with GPT-4 Turbo. The intervention reduced conspiracy belief by ~20%. The effect remained 2 months later, generalized across a wide range of conspiracy theories, and occurred even among participants with deeply entrenched beliefs. Although the dialogues focused on a single conspiracy, they nonetheless diminished belief in unrelated conspiracies and shifted conspiracy-related behavioral intentions. These findings suggest that many conspiracy theory believers can revise their views if presented with sufficiently compelling evidence.

And, oh yeah, "Gish Gallop?"

 

For some AI-related thoughts, I advise Shannon Vallor and Leslie Valiant.

 

    

Stay tuned... 

SEPT 21 UPDATE

There's a lot of detail in the Science Magazine "Artificial Arguments" related articles. Trying to restrain my skepticism until after I've plowed through all of it

apropos of thte pressing political timeline—45 days to the 2024 presidential election—any useful disinfo/ conspiracy-thinking mitigation applications they can sustain via replication will be on a much longer time scale.

I got on to this book this morning via Joyce Vance White's Substack:

Click

Deep into to it. Time is short.

A comment I left at Joyce's post:

Thanks for this. It goes almost verbatim to my long-standing concern about how 2024 may well play out. We are now only 45 days out from the election. I have downloaded Ari’s new book and begun reading it. I encourage everyone else to do it likewise. The prologue alone will blow your mind.
 

Remember that if Trump’s principal focus is denying 270 to throw it to the House, he starts out only needing Supreme Court backing of three justices should a supportive SCOTUS intervention be necessary; he doubtlessly already has Alito and Thomas in the bank. He will of course go through all the melodramatic motions of filing every lame objection in every venue possible just like he did in 2020, but all of that stuff to me is now just requisite sideshow noise. Denying 270 is the last-resort key. IMO it explains his lackluster final approach dilettante campaigning and bizarre ALL CAPS bleatings.

45 days. Ugh.

UPDATE

 

SEPT 25TH UPDATE

I finished Ari Berman's book. An excellent read.

UPDATE, BACK TO THE AI/LLM TOPIC

 

An Atlantic Monthly article led me to Quanta Magazine and to this imposingly bright young scholar:

 and

 

UPenn Phd, Computer & Information Science, Hopkins, Bachelors in ECON and Music Performance (she's a sax player!)

 

Stay tuned. More shortly...

_________