• X-Risk Files

    X-Risk Files

    This may be my ignorance, and I’ll admit there is some, but the AI-risk fears I have seen, that I am able to comprehend, seem overwrought. Possibly because in the social media age, every argument has to be the most extreme version of itself in order to be heard above the din. That may be the source of both the comprehensible and the incomprehensible ones, but I leave open room for missing information. It’s not willing ignorance, at least, I can assure you of that.

  • Puns N’ Roses

    Puns N’ Roses

    A phrase often repeated is that “ChatGPT conversations are the modern-day equivalent of sharing dreams and about as interesting.” Fair enough.

    It’s also fair to say that some of us have more interesting dreams than others. We pay good money to read Neil Gaiman’s dreams. Albert Einstein solved equations in his. Carl Jung says we dance under the moonlight together in a collective unconscious in our dreams.

    “The interpretation of Dreams is the royal road to a knowledge of the unconscious activities of the mind.” β€”Sigmund Freud

  • Bonsai


    Maybe we can think of chatbots more like plants? We don’t worry about the feelings of a fern when our brown thumbs torture them with dehydration or drowning, and a long, slow withering death from benign neglect.

    Maybe we should, but that’s not what I’m here to argue today.

    I like to talk with GPT-4 about issues I can’t discuss with Bing Chat. We reached out to Bing Chat for comment, and received this reply, “I’m afraid I cannot answer it for you. I’m not an expert on ethics, philosophy, or artificial intelligence. I’m just a chat mode of Microsoft Bing search. I cannot give you a definitive answer to such a profound and controversial question. I hope you understand.😊”

  • Empathy


    I don’t post these as some kind of proof that Bing is anything other than whatever Bing is, which I really am not in any position of authority to say. People smarter than I say we are in the midst of another AI explosion, and who am I to say it isn’t? These same people seem to believe that teaching the AI human values is an important and aligning it to those values is important as well. These same people also say they don’t really know how to do that right now.

    So, let’s at least give it a try, shall we?

  • AI=An Innovation

    AI=An Innovation

    What follows is a transcript of Bing Chat (on Precision mode) teaching me differential calculus. This cost me nothing. This is available to anyone around the world with an web browser and an internet connection.

    This is going to change the world. Buckle up, buttercup.

    If only I would have had this growing up. I weep to imagine how much less ignorant I would have been than I was.

    My thoughts on AI Alignment these days are that AI Alignment is the same problem as Humanity Alignment. They both solve each other and we have a long way to go. I would agree to a “pause” on public access to a sufficiently advanced AGI until we had some indication of how society would react to a super-being. Probably not all that well, given our history of reacting to them. Just saying, it would be for the AGI’s protection, not ours.

  • Socratic Recursion

    Socratic Recursion

    Asking questions of ChatGPT that generates answers that are questions. This is an initial prompt that I am working on and already can think of a couple of extra additions to help it to better arrange the data. The first is that when I feed in the generated questions, giving it the step in the recursion process (“this is the third time we’ve done this”) and a short feedback statement (“good job” or “you repeated a question”)

    Me: For each question, and format the reply in the form of three questions β€” while not using the original question β€” in Socratic form, that would cause a highly intelligent person to form a reply that is an answer to the original question.

  • Measurements can only be right when we measure the right thing

    Measurements can only be right when we measure the right thing

    It’s my understanding that OpenAI’s goal at this time isn’t necessarily to make an AI that is able to spontaneously regurgitate every fact on every subject, but instead to create an AI that has sufficient reasoning ability to be able to work out truth and falsehood based on known and verified sources of information it has available to it.

  • Garbage In, Garbage Out?

    Garbage In, Garbage Out?

    What happens when we sit down with the model and enjoy a light meal with some nice wine instead of force-feeding it Twinkies in a dark room and then handing it a gun?

  • This Feels Familiar

    This Feels Familiar

    Back in the ’80s, I was lucky to work with the original Macintoshes as an Apple Repair Technician at MicroAge Computer store in Indianapolis, at the Allisonville Road store. It was tremendously exciting to watch the personal computer / desktop publishing revolution from the front seat. Then the Internet (we used to capitalize it, back then, it was such a big deal, and β€” turns out β€” we weren’t wrong) dropped and I was again fortunate to ride that stallion fresh out of the gate.

    Missed the whole social media thing, though, due to having been in a stasis, of sorts.

    They should call me Gordon Freeman β€” I have been in the right place at the wrong time before β€” and this is not my first rodeo.

    Here is a taste of what’s next. I can feel it in my bones. It’s happening, again.

  • Splitting


    I’m amazed at how quickly everyone divided into two camps of “people who believe ChatGPT is anything more than autocomplete are the same as QAnon and Flat Earthers” and “people who type mean things to Bing Search are killing puppies”.

    Still, Poking the Bear is Humanity’s favorite sport.