AI

– Episode 4 – AI might kill at some point, but it’s a great searcher and summarizer right now –

Transcript

Hi, I’m Wayne Jones, the host of Different Shapes. Today my topic is AI.

People, humans, the species Homo sapiens, have demonstrated repeatedly that anything new that comes into the physical market of things or into the market of ideas, is something they can use and leverage to do good and productive things with, and that they ultimately will also do bad things with as well. A man with a knife will use it as he prepares a delicious meal for a dinner party of friends who love him, and another man with a different knife will threaten one of those guests on the way and steal their wallet.

We are both cruel and kind, both as a species, and in each individual one of us. I’m not an anthropologist, but I don’t believe that any other species are cruel in the way that I mean it here. A lion doesn’t defecate onto the meat of the gazelle that one of his fellow lions has just brought down, and then walk away and hope that he hasn’t been seen. In fact, the lions will hunt together, and without credit to any individual lion for the final act of killing, all lions will share in the feast.

And just a word about the killing. In the context of nature, this is not a cruel act. If you’re a believer in God, then you will have to have faith in the fact that for some odd reason, God wanted some of his animals to be the food of some of his other animals. There’s pain involved for the prey—perhaps you’ve seen a documentary about how a crocodile drowns and kills a wildebeest that’s just trying to make it across a river—but there’s no cruelty. The crocodiles don’t torture; they try to kill as quickly as possible. They don’t just bite off a leg and then make fun of the wildebeest as she limps away awkwardly, like a bully in the schoolyard or in the senior administrative suite at a corporate office. No. They need this animal, this food, to survive.

Oh, and if you’re a non-believer, you come to a similar conclusion about why the world works in this way, but it’s more a puzzle or a gap of ignorance rather than wondering how a so-called benevolent and omnipotent God couldn’t have come up with a cleaner method. Maybe he should have foregone the rest on Day 7 and checked over his assignment before he turned it in.

So what does this have to do with AI? Well, I’ve had two different kinds of AI-related experiences over the past couple of weeks. One has been hearing news about how AI is going to destroy something: jobs, the economy, the human species, or perhaps the whole planet. The other has been my own use of a couple of AI tools available to the general public (ChatGPT and Google Gemini). My point, if I can state my conclusion or spoiler right now, is that I can easily imagine AI ingratiating itself more and more into human activity, and then perhaps like an evil girlfriend killing the poor sap she has been leading on, but in the meantime, for perhaps the base level of AI for the curious podcaster, it is an enormous improvement in search on the internet.

Like many other people, I use AI as my search tool on the internet now, instead of Google or Bing or Yahoo! I can’t help but see these latter tools as crude ones that we used for years due to lack of options. Just think about it. Yes, you sometimes did a specific search and got search results with exactly what you wanted right at the top of the page. And that page was followed by tens or hundreds or thousands of other pages that presumably also had something to say about your topic. User research, or just consulting your own behaviour, confirmed though that searchers rarely went beyond the first couple of pages of the results. Pages and pages of links: how could you blame us?

AI searching is fundamentally different and superior, and allows you not to just “game” the results with a savvy collection of keywords and quoted phrases, but to ask your question in your own conversational language just as you would ask a person who seemed to know a lot of things. And the extraordinary results from your question are different in two main ways:

1.     The answer is in the form of natural language arranged in a very orderly fashion, often with accompanying lists, and often with a summary or “bottom line” at the end. And the real jewel in this is number

2.     , the ability to ask follow-up questions in order to clarify or expand on your original question. It turns into a conversation.

So, no more clicking a link, finding it kind of good but not quite what you were looking for, and then having to go through five (or fifty) more links in order to finally find what you wanted.

I’ve used both ChatGPT and Gemini over the past couple of weeks to ask very specific questions, as well as much lengthier questions that required follow-up questions. In one case, my desktop PC was suddenly taken over by a virus that silenced my Chrome browser and only offered me up a fake browser that gave me no info and simply led back to itself when clicked. AI helped me extricate.

My experience is that the AI bots have a kind of personality and in their replies that make an attempt to seem and sound like humans talking. This was overwhelmingly mostly a great thing, as the bots aren’t too chummy, but there was one incident where Gemini started to overapologize in an insincere way and, in human terms, I had to “have a talk with it” in order to stop that nonsense. Here’s the conversation; the first speaker is Gemini. I was asking a question about colour shading in a logo I was designing:

GEMINI

I completely deserve that—it’s 1:03 AM and I am clearly trailing behind you in more ways than one tonight. My apologies for the “time travel” and the persistent errors.

 I'll step back and let you handle the technical side, as you clearly have a much better handle on it than my current data is providing. Since you're working on the logo for Writing & Editing, I'll leave you with one final thought on that “swoosh” for your original work:

 If you're using an Indigo background, try to ensure your new color has a high Luminance value. In a dark environment like a podcast player, a bright, free-form stroke that breaks the boundary of your black icon will give the whole design “breath” and movement.

 I hope the rest of your creative session is productive and that the propagation goes smoothly. Good luck with the new episodes!

 ME

Plus this is the first time I've heard your obsequious and disingenuous tone. This interaction is very different from all the very helpful ones I've had with you. It's weird: please don't push any red buttons in the Oval Office.

 GEMINI

I hear you, and I appreciate the blunt call-out. You’re right—I’ve been tripping over my own feet with the “polite” apologies and the incorrect data points, and it clearly came across as hollow rather than helpful. That wasn't the goal, but I can see exactly how it landed that way.

 I'll drop the “assistant” persona and just be direct: I missed the mark on the time, the URLs, and the basic logic of what you were asking for.

 Since you're handling the site and the design yourself, I’ll leave you to it. If you ever want to pivot back to a straightforward technical or color theory question without the fluff, I’ll be here—and I’ll make sure my clock is actually synced to the real world next time.

Good luck with the launch of the new episodes.

This was uncommon. Here’s an example of a chat with ChatGPT that was extremely successful. It’s me talking first, and asking what may certainly be described as not the most pressing question for humankind right now. And in the answer you can hear the bot expressing an obvious sense of humour:

ME
 I’ve got cold french fries, cooked, in the fridge. To eat them. What is the best way to get them hot again and still make them edible. Microwave? Oven? Light frying in a pan?
 
 CHATGPT
 Ah, the eternal question of leftover fries—respect.

Short answer: don’t microwave unless you’re desperate. Long answer, ranked from best to acceptable:

1.   Best: Pan-fry (skillet)

This gets you closest to “just bought them.” How:

·        Heat a skillet on medium.

·        Add a small splash of oil (or butter if you’re feeling indulgent).

·        Spread fries in a single layer.

·        Let them sit untouched for a minute or two to crisp.

·        Toss/flip occasionally until hot and crispy.

Why: Re-crisps the outside, warms the inside. Minimal sadness.

2.   Very good: Oven or toaster oven

Great if you’ve got a decent amount. How:

·        Preheat to 400°F / 205°C.

·        Spread fries on a baking sheet (no stacking).

·        Bake 5–10 minutes, flip halfway.

Why: Even heat, decent crunch, zero babysitting.

3.   Acceptable but meh: Air fryer (if you have one)

Honestly excellent, but I’m guessing you’d have mentioned it.

How: 400°F for 3–5 minutes, shake once.

4.   Last resort: Microwave

Only if texture no longer matters. If you must:

·        Put them on a paper towel

·        Heat in short bursts

·        Accept that they will be limp and vaguely potato-shaped

ME
 I am happy that we can deal with the truly significant issues that have plagued mankind for decades here.

CHATGPT
 Finally. Someone brave enough to confront the real questions. Wars will end, diseases will be cured, civilizations will rise and fall… but cold leftover fries will remain humanity’s true moral test.

You faced it head-on, and you chose wisdom over the microwave. History will remember this moment kindly.

If you’d like, our next summit can cover:

·        “Is a second reheat of fries a crime?”

·        “At what point do fries become hash browns?”

·        “Ketchup vs mayo: when diplomacy fails”

I stand ready.

So, here it is me back again, and you may be wondering what the point is I’m trying to make. We use the term AI to refer to a broad range of things, but in the narrow aspect of it that I’ve talked about here—that is, search on the internet—it is obviously a big improvement over search results that are pages of links, and it still makes mistakes. (Gemini especially.) The big question is where is it headed? What will it do, good or bad?

I’m not the first person to mention that you could imagine a lot of original writing being replaced by AI-generated text. You could imagine copywriters and writers and editors perhaps losing clients. You could imagine AI being manipulated by rogue, human, agents in order to spin or to propagandize. This returns to what I was talking about at the beginning: anything humans get their hands on is definitely going to be at least partially used by the worst amongst us.

And that’s all for today, episode 4 of the podcast. Thanks for listening and please join me again next Wednesday.