Whatever The Hell That Is….

….it’s not thinking.

I write these posts and most everything else in Word, like most of the human race, at least in North America. Every time I start a document, and then every time I start a new line in a document, a symbol pops up in Word on the left hand margin letting me know that Copilot is ready and waiting to make my writing better and easier.

I cannot get rid of this annoying symbol, because the Word program I am using is not mine, I am just renting it from Microsoft, and I have very few choices about how it works. I expect before I die that Word will no longer give me the option to ignore Copilot, just as my Microsoft Windows OS no longer gives me many real choices about how my laptop operates. We who use computers -as in, almost everyone – are all just dancing to Microsoft’s tune, nowadays.

Well, I want to report here on a blog post I just came across on a website called Mind Matters, written by Gary Smith, an economist at Pomona College. It has nothing to do with Econ, trust me, and if you want to skip the middleman (me) and read his long-ish post yourself you can do that here.

Assuming you are still with me, I’ll give you the basic story and then some related commentary based on other stuff I’ve come across on the interweb thingie regarding LLMs (Large Language Models).

Smith is reporting on what happened when he put a question to four AI LLMs – Deepseek, Copilot, Gemini Flash and OpenAIo1.

Here’s the initial question, which Smith asked in two parts –

Gary:  I’d like to discuss a new version of tic-tac-toe called 90-degree tic-tac-toe.

The four chatbots all respond enthusiastically with some variation on – ‘sounds interesting, tell me more’

Gary: It is still a 3×3 grid but the grid is rotated 90-degrees to the right.

Smith says he edited the LLMs responses down a bit, as they do go on and on, and here are the four responses.

OpenAI:  Interesting concept! By rotating the grid 90 degrees to the right, you’re visually changing the perspective of the classic 3×3 Tic-Tac-Toe board. The gameplay itself might remain the same, but the rotated perspective could create a novel experience for players.

DeepSeek: Ah, I see! So in 90-degree tic-tac-toe, the 3×3 grid is rotated 90 degrees to the right….This could be a fun twist because the rotated grid might disorient players initially, making it harder to spot winning opportunities or block opponents.

Copilot:  That sounds like a fun twist on the classic game! By rotating the grid 90 degrees to the right, it gives players a fresh perspective and challenges them to think differently about their moves.

Gemini:  This rotation doesn’t change the core concept of getting three of your symbols in a row, but it does significantly alter the strategy and how players approach the game.

Philosophers have an accepted definition of ‘bullshitting’. One is honest when one says what they believe to be true, one is a liar when one says things they believe are false, and one is bullshitting when one is unconcerned about the truth or falsity of what one says.

LLMs bullshit, because it is all they can do. They have no idea what is true. One writer on AI that I read said that what comes out of an LLM is simply the result of a billion incredibly fast coin tosses. Not exactly right, I suspect, but close enough.

As for LLMs learning as they go, Smith goes on from here and drags more BS out of the LLMs. He asks them if they think it would be better to rotate the grid right or left, and they have definite opinions about that, as well as on the question of which direction of rotation will make the game more difficult for expert players.

Smith notes some key features of how LLMs generate answers –

They write hundreds of words when a dozen would do, they are always positive and enthusiastic (lots of !s), reluctant to say ‘I don’t know’, and they lack common sense.

His bottom line: ‘If you know the answer, you don’t need to ask an LLM; if you don’t know the answer, you can’t trust an LLM.’

All this made me realize something. LLMs are algorithms that rather faithfully replicate the behaviour of all the bad students I taught over the years.

I never asked painfully obvious questions about tic-tac-toe during my years in the trenches, but I almost always asked a question or two that could be answered with a few simple lines by any student who had studied. (I never asked multiple choice questions in my last 20 or so years of teaching.) And, there were always some number of students who gave long and ludicrous answers to those questions, because their lack of effort in the course meant they had no idea there was a simple clear answer. They were left trying to guess at words and phrases that would match up well with the words in the question. They were, in a word, bullshitting, in hopes that something they wrote would get them some points, in spite of their utter incomprehension.

I don’t suppose any company is likely to come out with a general use LLM called ‘BadStudent’, though.

A further (last, I promise) thought. I bet if you put Smith’s rotated tic-tac-toe question to a hundred second-year undergrads, some positive number of them would answer with bullshit not too dissimilar to what the LLMs spit out. However, those students would be doing that because they were not confident that they weren’t missing something about the question, perhaps reinforced by a lack of confidence in their own thinking. The LLMs have no ‘reasons’ for bullshitting, it’s just what they are designed to do.