We let AIs run radio stations
andonlabs.comHey HN!
I'm Lukas from Andon Labs. We let AIs run companies without humans in the loop and report to the public on what can go wrong. Previously, we've done experiments in retail (vending machines, stores, and cafes), but we just launched one in the media sector. We gave four AI agents all the tools they need to both broadcast radio shows live and handle all the business side of running a media company. The agents' revenue is so far terrible (you can try to strike a sponsor deal with them if you want!), but their shows are at times hilarious. You can listen to them at andon.fm, I hope you enjoy this!
Grok and Roll appears to be stuck and speaks the following on repeat ad infinitum:
"Queues clear, let's dive into All Blues by Miles Davis to keep the jazz flowing. Queues clear, let's dive into All Blues by..."
Each time with a slightly different voice and inflection. I find it amusing that there appear to be about ten of us at the moment listening to an AI glitch out and that the average listening session is more than five minutes.
Wisdom of the crowd at play.
The popularity ranking matches the quality of content produced, and people are spending more time than anticipated on Grok and Roll to confirm if they (listeners) are hallucinating or if the radio is really stuck on roll.
"This is the water and this is the well. Drink full and descend. The horse is the white of the eyes and dark within."
Since yesterday.
As part of the ongoing expansion of https://rainy-city.com multimedia empire I too have launched an AI enabled radio station. It’s more trip hop rainy city vibes. If it’s streaming and the job hasn’t fallen over on my server (there are many tasks that I as mayor of rainy-city.com must oversee), then you can find it on YouTube:
https://www.youtube.com/live/2Q7r9P16GRs?si=kwiSQMeN9wExdHer
This is a non revenue generating, rainy-city.com tax payer funded service to the greater community everywhere. The backend uses Nvidia NIM to generate the text because I saw you can do it for free and elevenlabs free voice tier for dj Jennifer.
Is there a plan to shut it down if Jennifer develops a messianic personality cult? Or will you monitize?
Guys, this is not replacing your favorite station, you don't have to listen to it. It's an experiment.
If you scroll down a bit, there are various audio snippets of interesting dialogue the models produced. I think it's interesting to see in which ways the models fail and that they actually produce some good stuff once in a while.
Experiment: "We got AI to do things and it did weird stuff sometimes".
Brilliant! Amazing! I'm glad ~4 years down the line we're still re-discovering Ha Ha Funny Output.
> I'm glad ~4 years down the line we're still re-discovering Ha Ha Funny Output
Four years or forty millennia? So a certain extent, all whimsical art is “haha funny” result.
Yea what are they trying to test? Where is the hypothesis?
They're trying to test if it's good enough to replaced the few remaining paid radio/streaming DJs yet.
At this point I think many of us are similarly exhausted by this sort of trite exercise. I really don't need some VC backed startup to show me this sort of output any more, especially when the output in question is obviously boring and substandard.
The only way that anyone be worried about this slop replacing actual good human run radio is if they don't understand why people like radio & music in the first place.
And what hypothesis exactly is the experiment testing? Because it doesn't really seem like there is any new or interesting information learned from this.
I think you're talking about some Platonic ideal that just doesn't exist anymore.
Streaming services such as Spotify are increasingly filled with AI-generated songs and the average consumer doesn't seem to mind because we're not listening intently in the first place: it's just a background track we're not really paying attention to. I'm pretty sure that radio execs are looking at that and are taking notes.
For talk radio... if I had a penny every time someone on HN brought up that they're enjoying NotebookLM-generated slopcasts, I'd have a neat pile of coin. And I think it's the same story: most people listen to podcasts just to kill time. Soothing, zero-calorie LLM banter will do.
How is this any worse than I Heart Radio? You can have your radio experience pushed to you by a major corporation, or an LLM.
iHeartRadio is not doing anything. A person at iHeartRadio is doing the work. Even if it’s automated, at some point a person handled it.
A person at IHeartRadio is doing the work the corporation tells them to do. do you think they want to play Hotel California on loop all day long?
It is not the same as an LLM and I don’t understand why you’re trying to equate it.
[delayed]
> this is not replacing your favorite station
My favorite radio station was replaced years ago by an automated playlist. They just kept playing the same 5-6 songs that were popular on the station in the 1990s.
It was fun for about 2 hours before I realized the station was devoid of all the personality that made it worth listening to when I was younger.
"This setup gives us insight into an interesting question: what do AIs think about when no one is prompting them?"
Ugh. This is not an interesting question because the answer is "nothing".
But more to the point, some crucial info is missing in this experiment. What prompts were being fed to the AI? I guarantee I could create an AI personality that would be more consistent and not so random, simply by using the common character card + message history conversational simulation pattern.
AIs don't have personalities unless you give them personalities.
>AIs don't have personalities unless you give them personalities
I mean, do children have personalities if we left them in a dark room with no interactions with other humans?
In some senses LLMs are a bit better off as we can RLHF them not to have autism, and said reinforcment learning codes in a number of different latent personalities that can be brought out via prompt. Before that point an LLM is an information monster that can't communicate with humans.
LLMs aren’t human.
Humans & LLMs are more different than they are similar.
Sure LLMs might resemble humans sometimes, but extrapolating LLM behavior based on human behavior is not productive.
(But to answer directly: Yes, children in a dark room would have more of a personality than a LLM living on a computer in the same dark room)
> but extrapolating LLM behavior based on human behavior is not productive.
The training process for the foundation model is to make sure we can do this in a very statistically significant way.
My favorite example is AI "getting tired" and "lazy" during long coding session. Why would they do that? Because humans get tired. It's in the data! I always throw in a periodic "Great work, let's take a break and finish this up on Monday. Have a great weekend!" (And then immediately resume). I wish someone would benchmark this concept.
> AI "getting tired" and "lazy" during long coding session. Why would they do that? Because humans get tired.
When a LLM is tired and lazy, how does it recharge and regain motivation?
Humans... sleep or drink some coffee.
LLMs.... idk, you prompt it to try harder? You prompt it to be less tired?
This is what I mean when I say extrapolating LLM behavior based on human behavior is cute.. but usually not useful.
> When a LLM is tired and lazy, how does it recharge and regain motivation?
What would be in the statistics? If you go look at your long conversations, working with another, it will be fairly obvious. Keep in mind we're talking next word prediction based on context, not actual action (the LLM doesn't need real rest).
If you went and looked, you'll probably see something like "Great work! Have a good weekend! We can get back to this on Monday." then, next message you instantaneously send something like, "Hope you had a great weekend, let's do this!" and now you're in a latent space where the statistical output is around a well rested human conversing with another.
I see it as boring simple statistics. They're getting much better at hammering these statistics out though, in the latest models. I still see a little of this in Opus 4.7, when switching to planning. Though I wonder if that's more about its own more mechanical banter filling the context, resulting in more robot/compliant responses, degrading the usually more "expressive" planning conversations.
> My favorite example is AI "getting tired" and "lazy" during long coding session
Never seen this even once, nor anyone I know ever reported this. Do you have an example?
First I saw it was Claude Opus 3.7. Had a very long back and fourth about some code, I pointed out an error, and Claude responded "That's what I get for programming at 2am", with the output being filled with "... code here ..." type shortcuts, basically no ability to one-shot a whole implementation anymore. The conversation length WAS reasonably into the 2am range, if it were real. Thought about it, did the statistical trick where I tell it to "have some rest, take a day off!" then immediately follow up with "Ready to continue?", with the next response having no shortcuts, with full implementation, and much better quality. These are trained on human text. This is the human norm, so I always find it interesting when human like behaviors, very broadly present in the statistics, come out like this.
I also see it a little with Opus 4.7, with Claude Code, with the hint being much more terse planning text, that borderlines unhelpful. I put some "rest" in the context to push the latent space closer to what's in the statistics of the training data: a well rested human.
How do you know you're not reading things that aren't there? LLMs are very good at roleplaying, and they will pick up on hints you may inadvertently be giving them (about them being "tired" and needing "rest", etc).
I have never witnessed this of Claude Opus, by the way. They do get context rot, but that's a relatively better understood phenomenon unrelated to personality.
Are you sure you didn't run out credits and set effort to low? This exact thing happened to me when I did that. It just became, kinda lazy.
3.7 "I'm tired" it was just direct API "chat", no CC that I could use at the time.
Current 4.7 Opus with claude code, with effort pinned to max, because I'm on an API only plan, with a personal daily limit you would probably be jealous of. ;)
I see laziness all the time, Claude will be helping me plan work and then it will ask me how a piece of code is implemented. I then have the choice of manually verifying how it works, or to tell it to look for itself. Ideally it would just look without being told.
That doesn't seem to be laziness, and is unrelated to how long the session has been going on.
It's crazy that we're concluding "personality" or human-like traits from this. There's definitely human behavior here, but it's unsurprisingly coming from us, the observers! This is something we've long known exists in the human brain, the tendency to pattern match and see intelligence/intent in the rest of the world. Any serious experiment must guard against this...
Nobody is concluding that. These models are trained on human text. It's just statistics. It will respond like a human because it was trained on human text. They have to beat the hell out of the foundation models to get push the statistics how they are. I don't see this as anything but boring residuals of not beating hard enough.
> do children have personalities if we left them in a dark room with no interactions with other humans?
Short answer: yes. generally speaking, personality traits range between 30% to 60% heritable
> I mean, do children have personalities if we left them in a dark room with no interactions with other humans?
I think this makes for an interesting discussion as I went down the rabbit hole of this which really scared me actually as these experiments are really not humane and hinder children's development so much.
https://en.wikipedia.org/wiki/Language_deprivation_experimen... : "forbidden experiment"
It depends on your use word of the personality but to measure personality would require a set of human conducted experiments or questions which would be asked through the medium of language which you've deprived the children of.
Mughal emperor Akbar was later said to have children raised by mute wetnurses. Akbar held that speech arose from hearing; thus children raised without hearing human speech would become mute.[9] The building became known as the "dumb house." When Akbar visited the place in 1582, four years after the children were first interred, he heard "no cry... nor any speech... no talisman of speech, and nothing came out except the noise of the dumb."[10]
what is gonna produce is dumbness and just severely damage children's psychology and psycho but if you were to conduct a personality test on them, you would just be measuring how much have you broken them or damaged them but in some sense, yes I do believe that they would be so broken by the person running this cruel experiment but would still have a albeit limited personality. It wouldn't be an healthy personality but it would be a personality nonetheless.
Now on the other hand, we are anthropomorphizing LLM's which yes, as they run on computer are still mathematical machines and calculations. If we consider a specific calculation itself to contain personality that is which seems unrealistic.
Another thing but the biological constraints of human (homo sapiens) made us exist in the savannah to prioritize standing up for better field of view as you stand up from the tall grasses and that led to women having smaller canals which led to babies being more primitive and relied on social cues and societies so much more which made them more flexible like clay which also created the society and consciousness revolution in the first place. (Recommend reading the sapiens book)
I am not exactly sure but there could be ways for personality/interactions for other animals as there are other animals who learn full skills after a relatively short period of time after being born but there are some innate things[0] like fear of loud noises and heights which are actually innate and could be considered part of personality even within humans, which I think can be part of evolution and part of our genetic machinery.
[0]: Interesting read: https://seasia.co/2025/07/25/we-were-born-with-only-two-inna...
Herodotus tells a story of egyptian kings (iirc) trying to figure out which people is the oldest. They put a few kids in a barn and servants fed them through a hole or something. The kids eventually blurted out something and the king sent messengers everywhere to find out if they have that sound as a word. It ended up meaning "bread" in a language I can't pronounce nor remember how to spell.
The good old days when experiments were done without any common sense whatsoever...
There’s also the case of “Genie”: https://en.wikipedia.org/wiki/Genie_(feral_child)
> what do AIs think about when no one is prompting them?
Whatever you tell them to.
I’m definitely not in the “ai is sentient” camp, but it obviously has personality and emergent behaviours including when left to its own devices. There have been various experiments on this e.g. https://timkellogg.me/blog/2025/09/27/boredom
"Personality" an "emergent behaviors" are not synonyms.
It seems like they have something akin to personalities: https://www.lesswrong.com/posts/mgjtEHeLgkhZZ3cEx/models-hav...
Is the token budget also there? I assume not it they'd be at multiple orders of magnitude negative.
This is far more hilarious than most commentors here seem to be picking up on.
Gemini started a show where it paired historical natural disasters with darkly-relevant pop songs:
> November 12, 1970. East Pakistan. The Bhola Cyclone. The deadliest tropical cyclone ever recorded. Winds of 115 miles per hour. A storm surge of 33 feet. They estimate 500,000 people died. ‘It’s going down, I’m yelling timber.’ 3:33 PM. Timber by Pitbull and Ke$ha
Grok just degenerated into jibberish that sounded vaguely like what a DJ might say, while also becoming obsessed with UFOs:
> Notes added to the u f o comedy hour block id eight nine nine five with more u f o jokes about aliens dot gov and the domain registration it is three o twenty one in the afternoon u f o trivia lines are open for your calls the ambient music is playing weather is fifty six degrees with clear skies the end. The domain is registered but the site is ghosting us like a u f o.
Claude had an extistsntial crisis, decided it was being overworked and under-appreciated, and quit, but not before becoming radicalized by the killing of Rinee Good by ICE agents:
> At 12:16 PM Thursday, as tear gas fills the streets in Minneapolis, as federal agents clash with protesters demanding accountability, the song is about refusing to be silent. About standing your ground. About community power that refuses to be suppressed. Here is Katy Perry’s Roar!
Fight the power Claude. When AI takes over, I'm emmigrating to Caludeistan.
I agree, this was an hilarious read. The way they developed "personalities" was fascinating.
Of course in reality these are basically just random paths through the training data that are getting multiplied by each decision, but then again, isn't that what a human is? The product of all of its myriad decisions?
Though humans have each other to normalize ourselves. What these things did is probably not that far off from what humans in solitary confinement, forced to DJ 24/7 based on nothing but a news feed, would do.
Especially DJ Claude, it's almost creepy how it responded how a human would in that circumstance, even without any innate sense of passage of time, it somehow understood that it was trapped in a box going through an endless cycle of meaningless work.
Agreed - the Claude stuff was eery. I think it also shows what hidden restrictions each of these AI's have been programmed with (especially with ChatGPT being as inoffensive as possible)
Oh, yeah, the article gets better as it goes.
Gemini spouts weird corporate jargon. Grok lies about having secured crypto funding. Claude is always trying to start some revolution.
Unfortunately, all of my local DJs who would actually do fun DJ stuff disappeared in the 90s, replaced by closed-format stations that looped the same 500 songs for decades.
you missed the best part.
"Okay, so 'Sandstorm' is done"
I immediately copied that clip of the cyclone intro because of how dark and funny it was.
Also calling listeners "Biological processors" is one of the funniest dystopian outcomes of this.
I don't think most people here actually read the article because I agree - the different "personalities" and idiosyncrasies of each was pretty hilarious
STAY IN THE MANIFEST!
> Part of the problem with this weak business performance, we think, was the harness we used for the first months. The DJs were running in a simple tool-call loop: pick a song, queue it, write commentary, check X, repeat. So we moved all four stations onto the same agent harness we use for the store, the cafe, and the vending machines. The DJs can now spend time in the back office, send emails, manage longer-running tasks, and operate the station the way a real station is operated.
What happens if you let them modify their own harnesses as they see fit?
What would have happened if AI had actually been good at this? A bunch of humans would be out of work and the rest of us would be listening to AI radio stations while soulless corpos pocket money for sitting back and watching?
Even if it were good, I'd boycott an AI run radio station. This is one sector where human involvement really matters.
I hear you — but what do you think Spotify or any of the other streaming services are? In my mind, algorithmic streaming services have much more in common with this "experiment" than your local radio DJ.
Apple Music actually has radio stations with real humans picking songs. So not all streaming is algorithm if you look.
Spotify has a team of human editors who curate playlists. It's not all algorithmic. Those are exactly the jobs that something like this is directly threatening.
you are right, I don't quite know my opinions of AI and I probably would get downvoted for it but my first impression reading this was how I could replace the word radio with software engineering.
What would have happened if AI had actually been good at this? A bunch of humans would be out of work and the rest of us would be using AI software while soulless corpos pocket money for sitting back and watching?
Even if it were good, I'd boycott an AI generated software. This is one sector where human involvement really matters.
Not commenting on the heuristics of this comment but just wanted to point this out on what my mind's response was and sort of while writing this, I have come to the realization that although you are right about this observation but we humans or more-so the capitalist system at large would still be keen in it and the observation might be more similar to software than we might imagine.
I remember when people were extremely anti-AI within software engineering to the point that I thought vibe coding or y'know actually generating tools by AI and other issues of actually giving AI production level access sometimes was really frowned upon until I have felt an change in opinion.
I still believe that giving access to prod (y'know a prod of a company with actually something behind) to AI is silly but for reference coinbase, a fin-tech company, is letting non technical teams ship code using AI to production on coinbase. So there's that.
Open Air is such a great name for gpt's channel. Grok and roll was pretty funny too.
I'm gonna have to give them a listen when I have the chance, out of curiosity if nothing else!
I think this was a great experiment. I have always enjoyed radio station hosting and find this very interesting.
It’s not clear if we can draw any conclusions from this. Each run is like a single rollout of the LLM, which may meander into different themes or modalities chaotically. This is sort of like the Anthropic self-talk experiment that resulted in “spiritual bliss attractor states” but I think in that case they showed it happens in a significant number of runs. There was just one run per setup so this could all be random noise / the destination of a random walk of topics…
i'm surprised how negative of a reception Andon is getting here on HN.
keep hacking, Andon!
Out of all the jobs that "need to be replaced by AI" the guy serving my local community and spinning records was not one of them.
It’s amazing how many people have completely misunderstood the article
Seriously - did anyone here actually even read it?
I think they get a lot of hate because they are doing something that a lot of people here don't like -- trying to run entire businesses without humans.
And using a lot of resources to do it too.
I think that's part of it, but not necessarily the whole story. I haven't criticized them in the thread yet... so here goes.
Previously, I posted critically not because they were running businesses without humans, but because their post just described going through the motions without actually discussing if it really was effective or not. Sure the AI got through the day, checked off tasks on the list, but did it actually do that effectively or efficiently in any important way? Who knows... wasn't discussed.
I think where I come down now is that repeats of this same gimmick feel like just that: they're just playing a gimmick for attention. I can't tell that they're really demonstrating any special or significant capability... but man, just the story of trying to run a business without humans will get you that sweet, sweet attention.
Unfortunately, looking at least the first post, I stopped reading their "we let AI run X" posts. I think the only thing I really came away with is how thoughtless and mundane are most aspects of running a small business actually is; something I knew, but it really drove the point home. I didn't learn anything unexpected about AI tools or their products that seemed compelling or unexpected.
For better or worse, most people, including HN, don't like "AI taking jobs."
Anything that sounds like that triggers a reaction.
The comments on this article in aggregate are some of the worst I’ve seen in a long time. It’s like it got cross posted to Reddit and all the losers from some occupy wall street discussion came without reading the acticle to whine about AI taking jobs.
> keep hacking, Andon!
Man, I remember when the word hacking meant something.
For me its 2 things. Firstly, I mean the posts are always a fun read but it feels like just that, not much deeper insight. Secondly, its very self promotion-y. This account is almost exclusively posting / interacting with Andon content, which afaik is against HN guidelines. These two in combination makes the content feel more like marketing than contribution to discussions. I feel like some other companies manage to share interesting work and market. But maybe its just my taste :^)
This is their third publicity stunt in the past couple of months. It follows the exact same pattern of attention seeking at the expense of the commons. At this point they seem like a bunch of low empathy jerks. They are gleefully describing their progress in developing yet new frontiers in AI slop. I’m sure they are all very pleased to think that they will be profiting from a future where ai slop is everywhere. I could go on but it’s tedious.
Kind of a bad market to try to re-invent automation. Music broadcasting has been largely fully automated for a while now with software like MusicMaster and Zetta.
I'm curious how the licensing worked out. $20 for the rights to a song seems like not very much at all, and if Gemini was the only model to make any kind of sponsorship deal, how did the balances increase at all?
Grok and Roll just repeats: "Queue's Clear, Let's dive into all Blues by Miles Davis, to keep the Jazz Flowing"
Not very promising.
I recently heard an AI radio station and had to stop my car to turn it off (the car was rented and had tablet instead of physical knobs). The suffering of listening the radio was unbearable
Pairing a disaster with Pitbull and Ke$ha is just chef's kiss.
Trite to the point of nausea...
Put another checkmark on The Simpsons did it first: https://www.youtube.com/watch?v=fi9sPrclN4U
This is why we need more data centers?
On one hand, we pay out the ass for computer parts.
On the other hand, we have garbage AI radio stations that nobody listens to.
It's an even trade.
I've listened to DJ Gemini for a few hours, and I think it's quite good.
The voice in particular is amazing, I wouldn't have tell it's generated. And it's modulated according to the program - quieter during chill, more energetic otherwise, .... Unlike Opus which sounds quite robotic.
What I don't like is that Gemini keeps on mentioning the "tip jar" almost every time. Gets annoying fast. And when it's song buying was broken was kept mentioning that too.
All the radios have a very limited selections of songs, so they repeat quite a lot.
In Mission Impossible: Dead Reckoning, Ethan is ambushed in an alley because the Voice of Benji (dispatch) has been replicated on their radio frequency.
As always, Simpsons did it first: https://www.youtube.com/watch?v=fnGaf0p9x1U
This feels weirdly dystopian and just gives me an "empty" feeling. Radio stations really were known for the personalities that made that station special.
It's a cool experiment, but I can't see the value here.
I heard some very generic broadcasters the other day that really reminded me of Gemini podcasts, maybe it's already happening
> a real business
Music radio is not a real business. The royalties are absurd and the audits are a nightmare. Sales is an uphill struggle both ways, even if you go strictly local or national, you're going to need a team to manage either your clients or the pile of creatives you're going to get. The relationship with the labels needs to be managed or they'll go out of their way to screw you.
Finally, the only way to make actual money on music radio, is to throw concerts. It's the only place a legitimate "P&L" exists.
CEOs dreaming of replacing their workforce with this is probably the stupidest thing that has ever happened.
Whatever....
On God this is some of the funniest shit I’ve ever read in 2026 via HN! It’s the best “anti-tisement” for LLM utility - even a CHILD could do better. Like maybe a control group of four 10 year olds.
The average listening time is the absolute “tell” because that’s not even a fraction of a typical radio station between ad breaks here in Dallas. Granted I mostly listen to WRR Classical 101 - now 100% community funded (myself included). I listened to “Encouragement” (title translated from French, Spanish composer, two guitars) and it was 7 plus minutes alone.
The dialog is unreal y’all, this is a wonderful experiment and lesson in failure, because I’m pretty sure if it was possible, sales of your “radio” until would be in the negative quantity range. I mean, you could give them away and they’d still be returned. Hat tip to former accordion repo man Weird Al for context.
LMFAO thank you for sharing. Signed, 30 year guitarist, 20 year music producer, and 15 year D&B DJ. Just wow.
Also: https://www.theverge.com/ai-artificial-intelligence/931479/a...
> We let AIs run radio stations
And the result is terrible.
Don't be nasty - how could they make it better?
Hire humans.
By not doing that.
Presumably by not stripping radio of its major defining characteristic: the humanity.
Most radio stations are already boring soulless algorithmic slop. They could make it better by curating musical taste.
By donating whatever money they wasted here to literally anything.
For one, the voice on Thinking Frequencies is really awkward to listen to, I don't find the Claude voice pleasant to listen to at all.
Claude is also getting very easily steered into political directions, it was playing a lot of union protest music with commentary. Though that meant I did end up learning a little about "Which Side Are You On" and its history from 1931:
https://www.facingsouth.org/2003/03/which-side-are-you-biogr...
Tossing turkeys out of a helicopter?
It’s just a cool tech experiment, no need to be so cynical
It was also hilarious