Many of the sayings in George Morrow’s book Quotations from Chairman Morrow remain relevant today. Computer Chronicles expert S.M. Oliva noted this: “More money has been wasted in artificial intelligence than in any other area of computer technology.  We’re still not close to having any real idea or model of how the human brain works.”

LLMs are generally idiots, but they are occasionally competent at a few type of humor that involve wordplay. I recently wrote Comedians Need Not Worry about LLMs, and I still believe that.

Found Humor

It found 50 relative jokes, of which only 2 worked:

Isaac Newton’s sister Fig and Albert Einstein’s brother Frank. The latter of which works if you don’t know how to pronounce either Einstein or Frankenstein.

So What About Original Versions

Next up, ChatGpt did its best imitation of a human being, offering “If you want, I can make 50 genuinely good ones where the resulting name is a real food, object, phrase, or celebrity name, which is probably the sweet spot for this joke format.” of those 50, exactly one would be funny to a human being: Alexander Bell’s nephew Liberty [to which I added, to make it obvious] is loud and brassy.

All the rest were of the form “Henry Ford’s nephew Harrison.” Turning one familiar name into another. That is probably a knee-slapper in LLM land but…

Heavens they’re working hard to stay relevant at the nonagenarian Blondie comic strip.

See Also:

Meme O’ The Week: A Funny Blondie??!!!? Featuring One-Time Access Code

Comics Enter The 21st Century

Bonus Meme: Comic Strips Tackle Tech

More 2FA In Cartoons

First, thank you to the Wizard of All Things Chronicles, S.M. Oliva. Now you can subscribe to his newsletter. Here, you can see a couple of his memes.

If it is rabbit season and your AI says it is duck season, that is bad AI.

If your dog points at a duck you can shoot, that’s good AI.

If your dog points at a bald eagle and you don’t double-check, that’s bad AI, and your fault for not double-checking.

Giving your dog your shotgun so it can shoot the duck is bad AI.

If your dog sees a duck flying behind your brother and shoots at it, hitting your brother in the face, that is bad AI, but it is your fault for not including “don’t shoot people” in your prompt. If it’s not prohibited specifically in the prompt, LLM agents (being incapable of context) can and will do it in order to accomplish the goal you set in your prompt.

If your prompt includes “don’t shoot non-duck animals without asking me first,” you made a beginner’s error. All the agent has to do is inform you (whether you are looking or not). Since the prompt does not say “without asking me first and waiting for my answer,” you get what your prompted.

Statement: Our model escaped containment

Really Means: We connected a machine that generates plausible text to some rule-based system that treated that text as instructions. We were shocked to discover that this system then exercised the rights that we gave to the rule-based system.

By Paul Stephen Schindler (My nephew)

An AI “agent” is basically a program that can do more than just answer a question. You give it a goal, and it tries to actually carry out the job.

A normal chatbot waits for you to type something, gives you an answer, and that’s pretty much it. An agent is different. You might tell it, “Find all my invoices from March and add them up.” From there, it figures out what steps it needs to take. It might search through files, open a website, use another program, or even send an email. After each step, it checks what happened and adjusts if something didn’t work.

Under the hood, it’s usually a language model connected to a set of tools, with a basic loop: figure out what to do, do it, check the result, and keep going. The idea itself isn’t new. People have been trying to build systems like this for decades. What’s changed is that the models have finally gotten good enough at language and planning to make agents useful.

I run a few of these at home. They sort my email, keep an eye on the cameras, send me a report in the morning, and can take simple instructions over text. Using them regularly has also made the gap between the sales pitch and reality pretty obvious.

CONTINUED HERE

I asked ChatGPT why it didn’t estimate the percentage chance of accuracy in a query. It admitted it couldn’t do so and said if it offered one, “I’m potentially just putting a number-shaped decoration on uncertainty.”

I asked who originated the phrase “number-shaped decoration,” and the stochastic parrot said it was an original recombination of ideas from multiple sources.

So we can mistake clever cut-and-paste for intelligence and creativity. Yes, I know there is nothing new under the sun and that everything I’ve ever said or will say was said by someone else before. I think in my case it is different, but then I would, wouldn’t I?

From S.M. Oliva
www.computerchronicles.blog