Bullshit and AI

In the comments to my previous post was a link to a paper on bullshit and intelligence.

“Navigating social systems efficiently is critical to our species. Humans appear endowed with a cognitive system that has formed to meet the unique challenges that emerge for highly social species. Bullshitting, communication characterised by an intent to be convincing or impressive without concern for truth, is ubiquitous within human societies. Across two studies (N = 1,017), we assess participants’ ability to produce satisfying and seemingly accurate bullshit as an honest signal of their intelligence. We find that bullshit ability is associated with an individual’s intelligence and individuals capable of producing more satisfying bullshit are judged by second-hand observers to be more intelligent. We interpret these results as adding evidence for intelligence being geared towards the navigation of social systems. The ability to produce satisfying bullshit may serve to assist individuals in negotiating their social world, both as an energetically efficient strategy for impressing others and as an honest signal of intelligence.”

Bullshit Ability as an Honest Signal of Intelligence: Martin Harry Turpin, Mane Kara-Yakoubian and Jennifer A. Stolz https://journals.sagepub.com/doi/full/10.1177/14747049211000317

I sort of hope that the whole paper is actually a complete fabrication in a power move where the authors prove their main point by producing a bullshit paper which paradoxically (because it demonstrates its own claim) isn’t actually bullshit. However, it seems genuine to me.

I’m predisposed to theories of intelligence that rest on humans being caught in an almost self-destructive arms race against human society as the main evolutionary pressure for intelligence. Humans got smart to cope with how other humans were getting smart.

There is an interesting point further down in the discussion of their results:

“Regardless of whether bullshit ability honestly signals one’s intelligence, of potentially greater importance is that skilled bullshit producers are perceived by others as highly intelligent. From the perspective of navigating social systems, being perceived as intelligent may be just as valuable to an agent as actually being intelligent, as this perception may afford one opportunities to obtain status and form relationships as well as have greater trust placed in their competence. To this point, we observed a strong positive association between bullshit ability and perceived intelligence.”

ibid

ChatGPT may not be intelligent but it is an absolute top-notch bullshitter: it states falsehoods with utter confidence and with great plausibility.

For example. Here is a prompt I just used in anticipation of ChatGPT producing some plausible but false bullshit: “Can you tell me about any essays written by the blogger Camestros Felapton?”

The first paragraph is a trap for the unwary but not BS exactly:

“Camestros Felapton is a pseudonym for an Australian science fiction author and blogger who is known for his humorous and insightful commentary on science fiction and related topics. He has written many essays on his blog, which covers a wide range of topics related to science fiction, fantasy, and popular culture.”

It’s really just a stock paragraph about science fiction blogging. However, look what it says next:

“One of his most popular essays is called “The Puppy-Free Hugo Award Voter’s Guide,” which was written in response to the Sad Puppies controversy surrounding the Hugo Awards in 2015. In the essay, Felapton provides a list of recommended works that he believes are worthy of consideration for the Hugo Awards, and encourages readers to vote for them.”

!!!! That is a popular essay but not one I wrote. It is an essay I have written about though. It was, of course, written by Deirdre Saoirse Moen https://deirdre.net/2015/the-puppy-free-hugo-award-voters-guide/ but less obviously it is not something I could have written because this blog and “Camestros Felapton” hadn’t started by April 4 2015.

Is it bullshit though or just a mistake?

Let’s look at the next example:

“Another notable essay by Felapton is called “The Ethics of Robot Sex,” which examines the moral and ethical implications of sexual relationships between humans and robots. In the essay, Felapton explores issues such as consent, power dynamics, and the nature of love and intimacy, and raises thought-provoking questions about the future of human-robot interactions.”

I’m a bit worried that ChatGPT is propostioning me now. This is quality bullshit. It’s a complete fabrication but utterly plausible. I was going to say “I’ve never written about robot sex” but…I actually feel like I need to double check.

I probably have to write that essay now.

Advertisement