One of the very earliest programs I painstakingly keyed into my first computer – a Commodore Vic-20 – was Eliza, the faux psychiatrist program that fooled non-coders into thinking the machine was actually thinking… and having a conversation with them. It did this by having a small array of prompts and then turning the conversation round on the unwitting subject, most commonly by asking “What makes you say that?” or “What makes you feel that way?”.
Watching its mostly-gullible users fall victim to the “it’s-god-in-a-box!” reaction was quite amusing, and spookily similar to what’s happening today, on a global scale, with AI.
Playing around with self-hosted Deepseek last night (anyone can run it locally on their own machine for free) brought back memories. The program is designed / intended to have exactly the same effect on the gullible, non-programmers. Its coders gave it a UI tone that is partly conspiratorial, partly obsequious… “I am your servant, oh master, command me…” It’s like having a tame genie.
The more socially-deprived amongst us will relish this transitory (and fake) thrill of power, the rest of us will grow weary of it quite quickly.
What’s it good at? Really, the same stuff that ChatGPT can do. On Pop-Up Submissions we had a workflow that included ChatGPT. The workflow was like this:
None of this is earth-shattering. It’s just using computing power to do a fairly menial task. It’s not going to change the world, but it enabled a fairly complex show to happen rather more smoothly.
Contrast this (sensible) use of the tech with the wildly overblown, and frankly scary, usage scenarios being trumpeted by the AI-boosters. Some of these folk openly gloat about AI’s ability to “crash”the wages paid to humans.
Well, if that’s what we all want… do we?
Watching its mostly-gullible users fall victim to the “it’s-god-in-a-box!” reaction was quite amusing, and spookily similar to what’s happening today, on a global scale, with AI.
Playing around with self-hosted Deepseek last night (anyone can run it locally on their own machine for free) brought back memories. The program is designed / intended to have exactly the same effect on the gullible, non-programmers. Its coders gave it a UI tone that is partly conspiratorial, partly obsequious… “I am your servant, oh master, command me…” It’s like having a tame genie.
The more socially-deprived amongst us will relish this transitory (and fake) thrill of power, the rest of us will grow weary of it quite quickly.
What’s it good at? Really, the same stuff that ChatGPT can do. On Pop-Up Submissions we had a workflow that included ChatGPT. The workflow was like this:
- Public submission made on our website by completing a form
- Form data goes to Google Sheets for processing
- GSheets send author biographical data to ChatGPT for sub-processing
- ChatGPT harmonizes style of author bio: puts into 3rd person; summarises to correct length for insertion into YouTube show notes; returns processed data back to GSheets from where the data is pulled into our wider workflow.
None of this is earth-shattering. It’s just using computing power to do a fairly menial task. It’s not going to change the world, but it enabled a fairly complex show to happen rather more smoothly.
Contrast this (sensible) use of the tech with the wildly overblown, and frankly scary, usage scenarios being trumpeted by the AI-boosters. Some of these folk openly gloat about AI’s ability to “crash”the wages paid to humans.
Well, if that’s what we all want… do we?