hpr-knowledge-base/hpr_transcripts/hpr4446.txt

Episode: 4446
Title: HPR4446: Calling on AI to resque HPR
Source: https://hub.hackerpublicradio.org/ccdn.php?filename=/eps/hpr4446/hpr4446.mp3
Transcribed: 2025-10-26 00:47:10

---

This is Hacker Public Radio episode 4446 for Monday the 18th of August 2025.
Today's show is entitled, calling on AI to reskage PR.
It is the 10th show of Troller Coaster, and is about 20 minutes long.
It carries a clean flag.
The summary is, LLM chatpots are going to explain us how to save Hacker Public Radio.
Hi everybody, it's Troller Coaster again.
I heard we're running low on podcasts, so I thought I might turn to Chippin and have
some fun with you guys, so I'm lazy.
So what did I do?
I just went to my favorite AI, and I asked for ideas, but I wasn't just lazy.
I was 1, 2, 3, 4, 5, 6 times lazy, because I asked Le Chat, that's Mistro, I asked ChatGPT,
I asked Deepseek, I asked Groc, I asked Gemini, and I asked Claude about how to save Hacker
Public Radio, and you know, they all had kind of the same ideas, and they also had kind
of different ideas.
So first things first, when I opened Le Chat, Mistro, I didn't need to log in.
He offered me to log in after question 2, so I could save my things, but he didn't
require me to log in.
For ChatGPT, I had to have an account.
For Deepseek, I needed an account.
For Groc, I didn't need an account, Gemini, yeah, for Gemini, I needed my Google account,
and Claude required me to log into.
Le Chat didn't show what model they were using, ChatGPT just had the basic ChatGPT for
three accounts, Deepseek didn't show any details either, Groc didn't show any details,
Gemini was using the 2.5 flash, and Claude was using Sonnet4, so this is some context
for what engine is being used.
So just to be clear, today is August 5th, and it's 9 o'clock in the morning in Belgium.
So let's go back to the question.
The first question I asked was, have you ever heard of Hacker Public Radio, and can you
tell me what it is?
So Mrol gave me a short, one paragraph, I'll answer a block.
ChatGPT, emoji galore, and that's the tone that we will have all over the place here,
nicely structured, and a summary in the end.
Basically, very easy to read, very easy on the brain.
Deepseek also has a nice explanation, not too short.
Groc is the only one that actually shows the URL in the content, the other ones have
it in the sources, and in the references, but they're hidden.
But Deepseek actually shows the URL for Hacker Public Radio in the text.
I found a bonus, Groc also gives us a short description in a block.
Gemini gives book points, Claude gives me a longer block of four paragraphs.
So that was a question, what is Hacker Public Radio?
The second question I asked was, you might have heard, they are running out of episodes.
In August, they will have to deplete their whole reserve queue if nothing happens.
What would be the consequences of them running out of shows?
Now, we all know, because it's getting set all over the place in the podcast, that
if there's a gap, Hacker Public Radio stops existing.
But hey, I have bad news for you guys.
You're wrong, because not a single butt set, HPR would stop.
They say there's going to be a gap.
There's going to be this option.
There's going to be a change.
But not a single AI said that Hacker Public Radio is going to stop.
So you're all wrong, guys, because AI knows the truth.
You're going to continue to exist, whatever happens.
You cannot stop AI agrees to throw lolololo.
I have to be some trolling in here anyway, don't you think so?
Come on, back to the topic.
So Mr. All had like a very business see approach.
It talks about things like disruption, organizational strain, restructuring efforts.
It makes no suggestions on how to fix it.
ChatGPT is all shiny, happy people, emoji.
They end with a part that's called not a doom and gloom, because there's an archive.
There's an IRC community.
It's a good thing, even if it stops.
But on a positive note, it also suggests some improved, some actions.
So it already suggests here to record an episode, to encourage your peers to record
an episode.
And it makes the suggestion to co-create or co-record thematic episodes, so like team
up and work on topics in a certain week.
It offers to help craft an episode idea.
So that's the end.
Can I help you make an episode?
That's a good thing, I think.
Then DeepSeek is the first one and it's opening sentences is, this is a concerning situation.
The hallmark daily would end, but other than that, it's loss of momentum, a threat to
viability, a threat to unique model.
The unique model, the threats are things that keep coming back and community is a very
important facet in the whole narrative from DeepSeek.
It keeps coming back all over the place.
It's talking about how it would miss out on capturing future insights as it has done
so greatly in the past.
In its score, it's positive because archive remains and also because recovery is possible.
And it ends with a very important conclusion.
The solution is the community.
Now who would have felt that?
And what can we do?
We can record an episode.
We can spread the word, we can listen and engage, for example, comment in forums.
We can donate.
That's what DeepSeek said.
So now going to Grok.
Grok is talking about a loss of listener engagement as one of the worst things that happens.
The extended periods without submission.
But it also talks about a minimal external impact because it's not a TV show.
It would only impact a niche audience, not impact a broader podcasting landscape.
I thought this was kind of bitter because I think it doesn't reflect the reality.
It talks about lacking corporate and political stakes as if this is a bad thing.
Okay, but it didn't make any suggestions on how we could help solve it at this point.
So Gemini was short.
Gemini talked about break in daily schedule as a consequence, a gap in the content and
risk of losing community.
So pretty minor, but no suggestions on how to resolve it.
That's it.
So anyway, so after all these observations, I did continue because I did say it seems
that if hacker public radio has a gap, the podcast will completely stop.
This will put everything in archive and the program stops to exist.
So it's crucial that we get more community engagement.
What techniques or tricks could we use to spark more user engagement and have the audience
create podcasts?
So really helped me create a call to action.
And again, I analyzed all responses.
So mistroll, I'm starting with mistroll because it's my pet peeve, it's your pee in one,
not saying it's the best one.
Anyway, it makes these suggestions.
Call for contributions, simplify the submission process, engage with the community, incentives
and recognition, workshops and tutorials, themed challenges and contests, supportive community,
showcase diversity, collaborations and partnerships, celebrate milestones, basically, you haven't
been doing any of these things, have you?
No, of course this is already what's happening and I want to thank you all for already
doing this.
So mistroll, chat GPT, what did you say?
First of all, emoji, emoji, emoji, emoji.
It talks about strategic framing, changing the narrative.
And despite the emoji, I have to say it had kind of a strong answer in my opinion.
It talked about really creative solutions, for example, it talked about an HPR relay challenge
where somebody creates a mini episode, three to five minutes, about some tiny topic.
And then in the end, challenges a friend do the same.
So basically you create like a chain of challenges and people take a friend to do the same.
It suggests making an HPR starter kit that's downloadable.
It also talks and this is where it actually shows some appreciation.
This is a bonus, encourage non-tech content because this is a part of what of the HPR charm.
Organize community podcast sprints, have like one moment in time where you rally everybody
and say, come on, let's do this together, start a community event and everybody recorded
their own local podcast.
I thought it was a fun idea.
She announced weekly themes and at supply templates.
Highlight first time contributors like you never did that before.
Collaborate with hackers, spaces and floss projects really call them out.
Reboot mailing list and master on presence to shout out to communities, write call-to-action
articles on related platforms, promote past episodes to spark inspiration, lower the
tech barrier and basically as a final note build belonging and not obligation.
Then the next one for suggestions I asked was of course a deep seek and this was a surprise
because here for the first time deep seek actually used emoji and so what were the suggestions
from deep seek, lower the barrier to entry but it set both technically and psychologically.
So be very explicit that a short episode is a good episode, doesn't have to be half
an hour or doesn't have to has to be well researched, just something out of the cuff is
just as good.
Then here I think the communism steps through a little bit, create urgency and scarcity.
Like ideas and provide structure, leverage community recognition, amplify call and target
outreach, technical and process tweaks in brackets if possible.
So that's what's also an appreciation of how well it's already done.
It gives a key messaging tone, it says be urgent but hopeful, be empowering, be demystifying
and be community centric.
And then it says all caps, create a highly visible, save HPR submission blitz, now that
was deep seek.
So then I went to Groc and I asked the same thing and here we had like a very business style
I found it was it creates seven, seven topic, seven points of action with each time actionable
steps and expected impact.
And what I didn't like very much about this and some other ones had a bit too, but here
it was the worst.
It says to grow the community, use your community.
It's like if what you're missing is the community, you can't use the community.
That's the problem.
Anyway, so it says streamline the contribution process, leverage community spaces for interaction,
demystifying contribution incentives, engage through targeted content calls, enhance promotion
and feedback loops, host live events and feedback loops, that's the one I kind of like.
So if you're on, pause them or somewhere, just announce where you will be and say that
you'll be interviewing people and asking questions and tell them where they have to be and invite
other, other, other podcasters to also do the same thing, make it community recordings.
Um, like a repurpose and highlight existing content, that's it, what it was last one.
So and then it creates like a plan with a short term, a midterm and a long term objectives
and ironically because I was talking about if it's in August, we run out of podcasts
and it creates a plan where you have a one to two month, three to six month and a six to
twelve month frame.
So it's like a two late and then it also offers metrics on how to kind of KPIs on how
you can decide if your project worked or not.
I didn't like it.
That went very much.
Anyway, then I asked Gemini and it was a really short answer.
It said simplify and demystify creation process, engage and empower existing community, broaden
appeal and reach create sense of shared ownership, nothing you hear.
Cloud said lower barrier to barrier to entry again, the five minute episodes, community motivation
and techniques, content inspiration strategies, leverage existing communities, gamification
and make it social.
So that's, I didn't go all too deep into these topics because I'll be sharing the links
to all of these and you can have a look at them in depth details of what they actually
are saying.
But I just wanted to have the different tones and different accents and different ideas
of how these language models, how their weights and biases, how their assumptions, how
their training leads to different approaches to the same problem.
So now let's go then to end everything I asked, can you give me five very different suggestions
for micro episodes that people could have because that people could make that have a very
low barrier to entry, just titles please and I'm going to read this to end the episode.
So what did Lysha suggest?
My favorite day gadget and why I love it, a quick tip for beginner coders, how I organized
my digital workspace, the most useful keyboard shortcut I know, a funny tech story from my
life.
Okay, that was Lichet.
Then chat GPT, what did you suggest?
Three open source tools I use every day, how I broke my system and what I learned, why
I still love command line email, the most unexpected thing I learned at my hacker space.
This is a funny one, explaining SSH to my grandmother.
Your public radio, for deep seek, what did you suggest me?
My one weird terminal trick that saved me yesterday, the five dollar or three tool that solved
my annoying problem, why I ditched popper app tool for something simpler.
My three step basic task without over complicating it, what I wish I knew before trying command
project or activity.
So these are five ideas for project school.
Grok, when I asked this question, it said, now you have to sign off with Grok three and
I couldn't continue and I didn't have the courage to set up an account at this point.
So Grok kind of stuck me in.
So I said bye bye my friends, don't have time anymore.
What did Google Gemini suggest?
My favorite command line trick, what's on my desk?
The tool I swear by, how I fixed it, the podcast that changed my mind.
And what did Cloud suggest?
My biggest tech mistake and what I learned, the one command tool that changed my workflow,
that's actually in my bin directory, a five minute tour of my desktop or home lab.
The weirdest bug I ever fixed.
When I go through all these suggestions, I think I'm kind of disappointed because they're
all really tacky, they're all really Linuxy, they're all really, yeah, they're not very,
not very diverse, they're not, not even that the one that suggested to also use non-technical
topics, didn't offer a technical, a non-technical topic.
I mean, come on, anyway, here are a few of my ideas.
If you want to make a podcast, you could make a podcast about your favorite online communities.
You could make a podcast, this is a classic, of course, over your favorite podcast recordings
or episodes, or maybe over funny videos you saw.
You could interview people on the train and ask them what they are listening on their
podcast player, so you could just walk up to random strangers and ask them, hey, can
I hear what you're listening and start a discussion about what they're listening if
is it a podcast, is it music, and spark a start there.
So that's just my idea.
Anyway, I hope these smart AI models will inspire us all to flood both the Q and the reserve
queue and, yeah, possibly, yeah, I mean, it won't stop because I said it.
So sorry Dave, sorry again, you're wrong.
It will stay and exist because not a single AI agreed with you.
Bye bye.
You have been listening to Hacker Public Radio at Hacker Public Radio does work.
Today's show was contributed by a HBR listener like yourself.
If you ever thought of recording a podcast, you click on our contribute link to find out
how easy it really is.
Hosting for HBR has been kindly provided by an onsthost.com, the internet archive and
our sings.net.
On this advice status, today's show is released under Creative Commons, Attribution 4.0
International License.