 Notice. New forum software under development. It's going to miss a few functions and look a bit ugly for a while, but I'm working on it full time now as the old forum was too unstable. Couple days, all good. If you notice any issues, please contact me. |
Forum Index : Microcontroller and PC projects : AI Is Evolving Fast
| Author | Message | ||||
Geoffg Guru  Joined: 06/06/2011 Location: AustraliaPosts: 3341 |
If you follow this forum you will know that Peter Mather has had great success in employing Anthropic's Claude to write code for himself. This is very good code, properly commented and it works well. A year or two ago I tried using AI to do the same, and it was a dismal failure, so I dismissed it as hype. But, what I did not realise is the speed at which AI is evolving. This was brought home to me when I ran into a quote from Dario Amodei, the CEO of Anthropic, who said that AI is now writing "much of the code" at Anthropic. So, AI is writing new versions of itself. No wonder AI is evolving so fast. This led me to an essay by Matt Shumer, an AI expert, about the impact of artificial intelligence. In it he explains, in a very down to earth manner, the speed of AI evolution and its impact on our future, particularly programming. I think that it should be required reading for everyone: https://shumer.dev/something-big-is-happening One of his recommendations is to skip the free, out of date, versions and subscribe to the latest. In his words " If you haven't tried AI in the last few months, what exists today would be unrecognizable to you". So I'm off to pay for the latest, I suspect that Peter is way ahead of me there. Geoff Geoff Graham - http://geoffg.net |
||||
TassyJim Guru Joined: 07/08/2011 Location: AustraliaPosts: 6461 |
I write programs for my own amusement and to try and keep the grey matter working. I have no need for rapid deployment. Unless it could auto-correct my rapidly deteriorating typing. I am impressed with the pace of Peters offerings but not ready to learn controlling the beast (yet) Most of my interactions with AI have been dismal failures. That is with the Google version. It tends to contradict the data sheets when describing the coding for various modules. When I searched for the reset procedure for my modem, it gave 3 options one was total rubbish, referring to a non existent reset button, the second was by dialing a magic number. Almost correct. It gave the wrong number. I should have gone to the user manual in the first place. Another query and it cited Facebook as it's source. It also managed to contradict itself by the second paragraph. Jim VK7JH MMedit |
||||
| Volhout Guru Joined: 05/03/2018 Location: NetherlandsPosts: 5719 |
Geoff, I pray AI is spiraling off in a good direction. It's growth is incredible, and it will be soon that no-one can keep up. It is already supporting criminals to become better at deceiving people. I am looking at it as if I am looking from afar to the beauty of a Nuclear explosion. Totally amazed. Staring at the beautiful colors, scared for the blast, and the nuclear fallout, knowing they will come. Not now, later... Call me religious, but are we worshipping a demon ? Volhout Edited 2026-02-24 17:54 by Volhout PicomiteVGA PETSCII ROBOTS |
||||
| robert.rozee Guru Joined: 31/12/2012 Location: New ZealandPosts: 2496 |
humans are the apex predator; we have reached every corner of this tiny ball of rock we live upon, we suck finite resources from deep underground, we fill the skied and the oceans with our trash, we groom and consume every lesser form of life to feed our appetites. and all the while we multiply. and multiply. are we not the demon, worshiping ourselves?  and now we feed AIs with every scrap of knowledge and experience we have amassed, we pump into the AIs massive quantities of energy and material resources. all in the desperate hope that AI will solve our own problems - overcrowding and the finite resources of a tiny ball of rock. the only 'success' that lays ahead is if the AIs surpass our failings, and become the new apex predator. then our own purpose becomes but to feed its appetites... until we are no longer needed.  cheers, rob :-) Edited 2026-02-24 18:19 by robert.rozee |
||||
| matherp Guru Joined: 11/12/2012 Location: United KingdomPosts: 10982 |
Geoff I now do all my development in VSCode. This has an AI extension "GitHub Copilot Chat". You pay a monthly subscription and get access to a wide range of different models - see below. You can use any of them within a certain "budget" each month. The fact that the AI is in VSCode doesn't limit its use. I've used it for things like creating 3D designs in OpenSCAD as well as programming for multiple environments. The integration with VSCode is great and one technique I've used and found helpful is to get one model to check the work of another. I predominantly use Clause Opus 4.6 which is expensive but the best IMHO and use GPT-5.1-Codex-Max to check the work. Note that using VSCode can get in a mess if you try and use a single installed version for multiple environments but you can create multiple separate VSCode environments and dedicate each one to a different use. Each of these will have full access to the AI once you load the extension in each case.   |
||||
Gizmo Admin Group  Joined: 05/06/2004 Location: AustraliaPosts: 5169 |
I've been a paying user of ChatGPT for about a year, its on version 5.2 now. I use it every day, and its starting to save a lot of development time. I like how it has learnt to understand my coding style. If I was a software developer with no other real life experience, I would be worried. I cant see a future for a dedicated traditional coder. But I'm not a coder, I'm a business owner with lots of experience in the industries I supply software to. ChatGPT has improved my productivity, but only because I know what to ask for and I have a understanding of how code works. Thats the future as I see it, we'll be the middleman, we talk with the customer, then work with chatGPT to write our code, then we test and refine. What is concerning is the number of fake images and videos been generated by ChatGPT. Facebook is full of them now, its all just fake news, but unfortunately many people cant tell the difference. Glenn The best time to plant a tree was twenty years ago, the second best time is right now. JAQ |
||||
| zeitfest Guru Joined: 31/07/2019 Location: AustraliaPosts: 656 |
Chances are, AI is already replacing the customers operations, or at least consolidating multiple businesses in that area. |
||||
| PhenixRising Guru Joined: 07/11/2023 Location: United KingdomPosts: 1740 |
Freaky....I just came to TBS to look for what matherp recently recommended. I couldn't remember the thread but lo and behold, I find this   |
||||
| PhenixRising Guru Joined: 07/11/2023 Location: United KingdomPosts: 1740 |
This is where I was going with my rant on Lizby's thread that was shut down Trojan horse: It used to be "it's true, it was on the TV" Then: "it's true, it was on the internet" Now, because LLMs are becoming incredibly productive, people will believe everything that they tell them. I recently came across a video where some guy asked for an example of something abhorrent in the Bible and it came up with lots of examples. He then asked the same thing about the Qur'an.....lots of examples. He then asked the same about the Torah but he had to keep pushing and the worst he got was "well some people might be offended by...." I forget what it was but it was a bunch of nothingness. You believe that no-one is steering this thing? |
||||
| JohnS Guru Joined: 18/11/2011 Location: United KingdomPosts: 4254 |
Pretty much. It needs to be using morals, ethics, humanism, etc - but isn't. Not, I suppose, a topic for this forum. John |
||||
| Volhout Guru Joined: 05/03/2018 Location: NetherlandsPosts: 5719 |
Ok, where do I share my thoughts if not on this forum. Maybe we should talk about it here, and share thoughts while we still can. Of coarse there is a lot of fake info on the web, that polarizes discussions. But, different from social media, forums can be the only uncensored media left (except our moderator). And, I am worried. Can't express myself here ? Volhout PicomiteVGA PETSCII ROBOTS |
||||
Mixtel90 Guru Joined: 05/10/2019 Location: United KingdomPosts: 8582 |
I completely agree about forums. Social media and, in fact most media, are often "poisoned" by false information. Forums are some of the few places where it is difficult for even AI-assisted bad-players to spread it. I think we could probably have a separate thread to vent off about AI for a while. :) Mick Zilog Inside! nascom.info for Nascom & Gemini Preliminary MMBasic docs & my PCB designs |
||||
| Volhout Guru Joined: 05/03/2018 Location: NetherlandsPosts: 5719 |
Mick, This is a special thread, started by Geoff, with a link to a web based article about the terrifying speed at which AI is developing. With a warning that everyone who's work is behind a computer should either quickly adopt (to gain some time) or be obsolete withing few years. This is THE thread. Volhout Edited 2026-02-24 23:08 by Volhout PicomiteVGA PETSCII ROBOTS |
||||
| Bleep Guru Joined: 09/01/2022 Location: United KingdomPosts: 735 |
There's a new 3 part series on AI fronted by Hannah Fry in the UK, on BBC2 Monday evening, I've recorded the first one, but they'll be on the IPlayer. This is an interview she did with New Scientist on the subject. I think Software is very much at the forefront, because it's relatively easy to be able to generate & test code, so yes coders days are numbered, but design and proper testing will come to the fore. But I write code as a challenge, for enjoyment, so it won't stop me coding. :-) Edited 2026-02-24 23:29 by Bleep |
||||
| robert.rozee Guru Joined: 31/12/2012 Location: New ZealandPosts: 2496 |
i am absolutely sure that NO ONE is steering, or in any other way controlling the progress of AI. this is, in good part, because we (humans) have already been left behind by the rate of progress. enabling this progress is the same simple free-market economic mechanisms that have dominated our society for many decades - it is a 'free wheeling' machine that propels itself at maximal speed without intervention. the interesting part is that, as humans we consider ourselves as moral, loving, on the whole 'all round decent' folks; we each have innate concepts of right and wrong. we don't know why or how we are like this, but (almost) every human being is this way. while an AI can be created in such a way that it displays behavior 'as if' it is the same as us - it can "act like" us - the troubling question becomes: is "act like" the same as "is"? an AI can certainly be created that will "act like" it is moral, to to extent that it is impossible for us to tell otherwise. but is this 'good enough' to say that the AI is moral? and can we trust such a thing with our existence? cheers, rob :-) Edited 2026-02-25 01:00 by robert.rozee |
||||
| PhenixRising Guru Joined: 07/11/2023 Location: United KingdomPosts: 1740 |
Have you seen the chats where one of the given rules is to respond with a code word when it is being instructed to give an untrue response? |
||||
LeoNicolas Guru Joined: 07/10/2020 Location: CanadaPosts: 563 |
I've been using AI professionally for a while. Like matherp, I'm using VS Code with multiple models, and the resulting code is great, including the unit tests created by de AI and used by the AI to test the generated code. It is not perfect. It requires a lot of guidance and instructions, some times with specific technical details to make it generate what I want, but the productivity it's astonishing. MMBasic is my playground. Basic is a nostalgic language for me, it was my first experiences as a kid programming computers. I'm not using AI to program in basic, I still like to write code, especially games, and the use of AI here will ruin my fun |
||||
| lizby Guru Joined: 17/05/2016 Location: United StatesPosts: 3621 |
I subscribe to ChatGPT, Claude, and Gemini ($20/mo version). I've had several all-day, every day sessions developing a pretty capable relational database in MMBasic for the PicoMite2, PicoDB, and now, a vibe virtual model railroad controller to exercise the database (presentation in time). It's mostly great fun, with productivity in stretches 6-8 times my own best 40 years ago. I'm mostly using Gemini with no wrapper for coding, and the others for various aspects of the project or for random queries. I am not close to a current state-of-the-art level, but would say I'm presently a fairly heavy user for a nearly-80 27-years-retired programmer. Debugging can be difficult--but it can also be a breeze, with Gemini quickly saying, "I see the problem" and providing a solution. As a bonus, Gemini produces documentation in minutes--a task I'm sure almost every programmer is glad to pass on. Yesterday was particularly satisfying--I started around 11 with a layout issue involving resizing everything that I wouldn't even have attempted on my own, and when I quit at 6:40pm, it was all worked out--along with several bug fixes and other upgrades along the way. In addition to the technical aspects, I find the conversational experience to be remarkable. After the first day on the model railroad project, I quoted the "Once I built a railroad" lyrics ("Buddy can you spare a dime"). Gemini immediately came back with the author of the song, of whom I'd never heard. It's understanding of obscure hardware and software is also notable. While I am an enthusiastic user, I am in no way comfortable with where this may be leading. Some quotes: "In a few years, humans will no longer be the smartest entities on the planet"; "The AI you use today is the least capable it will ever be"; "Intelligence wants to be free (as in "free beer"); curiosity no longer has a cost". We face a future where it is possible that humans will at best be the treasured pets of AI masters. I do not have confidence that political leaders in the U.S. and China and elsewhere are doing everything necessary to assure that humans remain in control. ~ Edited 2026-02-25 00:59 by lizby PicoMite, Armmite F4, SensorKits, MMBasic Hardware, Games, etc. on fruitoftheshed |
||||
| JohnS Guru Joined: 18/11/2011 Location: United KingdomPosts: 4254 |
When I posted that this wasn't the place it was only that I thought Gizmo/moderator(s) didn't want it. Seems we're OK so far :) John |
||||
PeteCotton Guru Joined: 13/08/2020 Location: CanadaPosts: 610 |
Very interesting points from everyone. I was going to recommend the same thing. It looks very even handed and well thought out. I would add a couple of pennies. I think that Ai might be the death of social media as we know it (and that's a good thing). I can already see fake bot accounts spamming the Facebook pages that I frequent. They're trying to build up a "back story" so they look legit when they start trying to sell you car detailing or furnace cleaning services (two huge scam areas here in Canada). Ultimately, if Facebook is full of these self-serving bots, then (hopefully) people will look elsewhere for the human connection. Secondly, I'm not concerned that AI will try to rule us. Brains are only a small part of what motivates us as human beings (and I'm not even touching upon the point that AI is not really intelligent). But our desires are driven by physical bodies (hormones, warmth, food etc.). Our survival instinct comes from millions of years of evolution, as does our sex drive (which arguably is the motivation for us to impress/succeed). An AI (as we know it today) has none of these drivers. It doesn't care if you switch if off tomorrow. It doesn't "want" anything. As a bonus third penny - I've been using Copilot a lot within Microsoft Visual Studio. It is great at producing reams of code - but the bit I think that is less impressive, but more helpful, is that the code suggestions are now intuitively guessed. For example: If I create a function called "DeleteRecordFromAList" and I have a List called AList, then it will suggest a function that includes a for-loop (you can't modify the list with "foreach"), iterate from last to first (so that the index doesn't get screwed up), and delete the matching item from the list. I hit the tab button - and pow! - there's 6 lines of code that I didn't need to type in. You do that twenty or thirty times in your day and your productivity sky rockets. Edited 2026-02-25 02:33 by PeteCotton |
||||
 Print this page |
 |
The Back Shed's forum code is written, and hosted, in Australia. | © JAQ Software 2026 |
