on point
3 Likes
I used Copilot quite a bit at my last job, and the improvements between Microsoft’s free tier, GPT4 (cheaper tokens), and GPT5 were mind-blowing. Free tier could sort of string together sentences that made sense; GPT5 could analyze and compare multiple documents, provide genuine insights, and develop a compelling executive summary. GPT4 took several revisions to develop a ~20 line Visual Basic macro that actually worked; GPT5 usually nailed it on the first try.
My current employer uses ChatGPT and I’m interested to see whether GPT6 is notably better for my use cases. But I will say that, even with the newer models, I never felt comfortable using its output without modifications. There were language patterns and formatting tendencies that felt natural enough in short bursts but started to feel very “synthetic” over the course of a few paragraphs. My cousin is a software developer and he has the same experience — the more you know about the thing you’re asking AI to do, the less impressive the results.
I wonder how long before that’s no longer the case! And RE: images and video, I can still usually tell, but it’s a lot harder to pick out than it was even a year ago…
4 Likes
Please bear with me, as the following story has a moral at the end…
I keep my educational videos on my public website. The website looks like it’s from 1993…a bunch of tables, displaying static content and links to my videos. The main page of the website is getting rather long. The other day, I started playing around with some alternate methods of displaying the tables. The AI summary to my general query showed some JS code for displaying the table headers only but toggling the body of the table on click.
At this point, I feel a bit yucky, because my understanding the JS script was minimal; I understood it just well enough to implement the appropriate class and id stuff in the .html in order to make the JS work correctly.
I set my cognitive dissonance aside, and said problem solved, cool! But now I can no longer perform a Control-F search on the hidden content. So, I search for answers to that problem, and I am presented with another snippet of JS code. But it isn’t compatible with the JS function that collapsed the tables in the first place.
Now, my guess is, at this point, any AI optimist would explain to me that I should merely include the collapse function in my query about the search function. I do believe using that approach would produce some sort of relevant solution. Continually feed the preexisting context into the AI. For lack of a better term, I’ll call this method the recursively share all your progress method of AI-enabled software development.
What I suspect, and this is open to disagreement or clarification, is that the recursively share all your progress method virtually guarantees that you are “over your head”, that you have “lost focus”. And I include computer experts with their self-professed “chops” into this group. Ignorance grows faster than whatever knowledge is gained.
Now, if results are primarily what you’re after, I’m sure you’ll get there with AI, but not before the AI data centers have gulped massive amounts of electricity and water. And you can’t wean yourself off AI, because of the recursively share all your progress approach you’re stuck with. AI companies win in the end.
AI apologists frequently refer to AI as just a “tool”. I don’t consider that analogy to be apt, because the traditional idea of a tool is being in the hand, controlled by the human. More like a tool that has been dropped into a black box. Humans need to understand how “out of our hands” AI really is.
8 Likes
I also think “just a tool” in its strictest sense doesn’t fit AI.
Sure, it can be used as a tool as well, but it has more capacity than that.
I would rather compare it to something like the element of fire. Sure, fire can be used as a powerful tool, but it can also cause disasters on its own.
To me, cat was already out of the bag as the algorithms (youtube & spotify recs, facebook feed) started making their way into everyday life. They are a black box all the same. Sure, they can be used as a “tool” but if let unchecked, they start distorting things.
What I see AI realistically achieving is that it allows non-professionals to create mock-ups of design, apps and websites. Before AI the barrier was higher and you needed some basic skills. But in the end the results are still mockups and don’t really stand up to deeper inspection.
5 Likes
Looks like the original OP1 is getting a lot of new stuff, firmware hacks for it have been around since 2016, but some of the new stuff being added is pretty great. The crappy weird TE envelope on the sampler which everyone hated has been replaced with adsr per key, tape has undo, lots of new fx and graphic view of waveforms on tape, and lots of other stuff that makes it a way better and more useful instrument.
1 Like
Yep, exciting stuff!
Unfortunately, seems like the OP-XY will be unlikely to get the same treatment… doesn’t have the same vulnerabilities.
Maybe by voltage-glitching or decapping a processor… but not with AI tools available now, anyway.
Never say never though 
Considering how the OG OP1 was a deeply personal project for Jesper, wonder if the thought of peeps changing his beloved synth/ethos is one the driving forces that lead to the recent we aint making no more synths talk.
Like if yall not going to appreciate my cooking, then i aint cooking any sunday dinner from this point forward steelo.
HOMER CARS FOR EVERYONE
(Shower thoughts)
5 Likes
I don’t think so, since the OP1 original has had alt firmwares for 10 years now, initially just unlocking hidden synths and fx already in the firmware.
1 Like
Yes, but that alt firmware only exposed things that were already there, made, and scrapped by TE.
Might be a different feeling when others tamper with your creations and mess up the intended ethos, but who knows.
We all graspin straws up in here
2 Likes
I have no actual facts to share… but that seems very far-fetched to me.
It’s more likely to be “we can make more money by doing this other thing”.
2 Likes
Could be, like i said in my post shower thoughts!
In the same regard, i dont think they operate conventionally in the normal businesses sense imo, feel like they lead with their heart, and i have read many interviews with Jesper where he talks about how deeply personal his vision was for the OG OP1, so was just a thought i had when seeing the op 1 being the most recent hackjob.
4 Likes
I think, actually, that forward thinking companies will think this:
How can we make this work in our favour?
It’s a lot of free labour, free advertising, extra purchasing incentives for people on the fence about buying.
I had zero interest in buying a Move. I could see it was great, but didn’t need or want it.
Schwung schwung it for me. It was the ‘push’ I needed, and I was Able to (n) purchase. I couldn’t Live without it (I’ll stop now 
)
At least one sale there… I’m sure others have done the same.
No matter how many people on forums shout ‘hOm3rCAr’ (and that’s no offense intended… just getting bored of seeing that term), smart companies are going to use this trend to sell more synths.
‘Homercar’ is used incorrectly, too, imo - ‘The Homer’ had a lot of unnecessary stuff, like 3 horns playing - 'La Cucaracha, some of these firmwares are adding stuff most people would consider essential.
Two concrete examples, Deluge getting a compressor, OP-1 the ADSR.
Sure, there’ll be Homercars happening. It’s inevitable.
But, I’m sure most people here will be able to discern that, and choose for themselves, right?
EDIT - I’ll give a little example (of how companies can make this work):
Minilogue XD. Okay, not whole firmware there, but the range of FX you can add, for free or cheap, is amazing.
Perhaps we’ll see more ‘plug-in-able synths/grooveboxes’, with FX or oscillators available.
2 Likes
You stopped just in time, my coffee had just evacuated my nostrils.
1 Like
I wonder if we’re in the Napster phase of the AI-DIY thing and soon we’ll be in the Spotify phase where a big company will monopolise all these mods.
7 Likes
splice cant wait to get a piece of these tokens
3 Likes
Had ducked out of here, but just one more for the road. I saw this earlier in the week and keep coming back to it. One of the most truly pathetic - as in pitiful - things I’ve ever seen in my entire life.
Just absolutely mind-blowing to me how LLM Brain invariably coincides with obviously not understanding anything about the human experience, the world, creativity, love or joy. The brainworms have brainworms. The kind of people who poured billions into the Metaverse assuming that this was obviously a great idea that people would love.
This guy was very senior at Musk’s twitter until a couple of weeks ago and I think presided over the explosion of custom CSAM generators - another great example of how AI is here and we’d better get used to it.
So not just picking on some random dude - a small window we can look through and get a glimpse of how the top echelons of the tech elite see the world. The people who are responsible for determining the shape of this technology. Many such cases!
6 Likes
I’d be all for it, if it were Pong.
yeah tbh i would love every truther to contend with the reality we live in where the wealthy and powerful people enabling their magical enhanced productivity buddy are also saying “oh yeah make as much CSAM as you can with this sucker”
2 Likes
yikes. sadly i have a friend who says stuff kinda like this. in the most optimistic scenario, that this is possible within a decade, no one would want to play this guy’s life. they’d be burning electricity created their own version. it really blows my mind how many people just gloss over the environmental effects of their actions/lives. someone might say we’ll figure out power, but like let’s do that first!?!? get power into rural africa so people don’t have to walk miles to collect water.
also i searched CSAM and a terrible category of stuff came up. child abuse material. is that what you’re referring to?
1 Like
yup. thats exactly what it is
from an organization preventing child abuse: AI-CSAM Ruling: What the Seventh Circuit’s Anderegg Decision Means for Child Safety - Enough Abuse