Tag: LLM

Technology is never just a tool

Let’s be clear on the background mess, before the personal attacks start, this is not about individuals. It is about patterns, systems and ideas. The danger is that criticism becomes an #adHominem argument – “you just dislike this because…” – instead of looking at the actual structures being discussed.

The point I am making is that parts of dead #postmodern thinking have ended up embedded inside #neoliberal culture: fragmentation, individual identity, endless discourse and difficulty building any shared collective action. That does not mean every idea, person or piece of work in those spaces is the same, it means we need to look at how ideas interact with power.

The question is – What helps us build collective capacity in a time of #climatechaos, inequality and the #dotcons mess? What creates commons? What creates shared action? This is the conversation.

So with that in mind lets look at the major problem with the #dotcons attention economy the advertising model. The platform logic and the attention economy are now becoming harder to simply ignore. For most of mass media history, the commercial transformation of media was hidden behind a layer of journalism, culture and public value. The advertising model was presented as simply a way to pay for content. Platforms were presented as neutral spaces for communication. Algorithms were presented as tools to help people discover what mattered.

But the #dotcons direction has now stripped this bare – the direction has become clearer, the media landscape looks less like a place for shared knowledge and more like a shopping catalogue with occasional content attached. The focus is no longer even the fig leaf of informing people, connecting communities or building public understanding. The naked goal is simple – more clicks, more engagement, more time captured, more data collected and more consumption encouraged. This is the logic of the #dotcons.

The problem with this #deathcult worshipping mess is not only that companies make money. The deeper problem is that the structures built around making money reshape our culture itself. When attention becomes the product, everything starts being measured through extraction. A story is only valuable because it generates traffic – A person is only valuable because they generate data – A community is valuable because it creates engagement – A conversation is valuable because it keeps people inside the platforms. Any, social value gets pushed aside.

The original #openweb grew from a different idea. People built websites, forums, mailing lists, software projects and communities because they wanted to share, collaborate and create. The value was not only in the information produced, the value was in the surrounding relationships. People corrected each other, developed trust, knowledge was maintained collectively.

The internet worked because there was social infrastructure around the technical infrastructure. The mess we made, was thinking that communication could simply be handed over to commercial platforms without catastrophic changing the nature of communication itself. A platform is not just a tool, it comes with incentives, has owners, rules, a business model. When every space becomes a marketplace, the culture changes.

The mess we have made is that extraction replaces participation, the #dotcons path works by turning human activity into resources. People create, platforms capture. Communities produce culture, companies monetise attention. That extraction eventually damages the thing being extracted from, creators become exhausted, communities fragmented, trust declines as people become audiences instead of participants.

The internet becomes full of “content”, but much poorer in meaning, more information does not automatically create more knowledge, more communication does not automatically create better communities, without care, context and collective responsibility, abundance becomes noise. To compost this mess we have made in the media tech path – the question is not “How do we get more people producing?” The question is “How do we build systems where what people produce strengthens the commons instead of feeding extraction?”

The fashionable people of #AI are pushing at changing the scale of content creation, lowering barriers to producing books, apps, music, legal documents and academic papers. Thus, “output” is exploding. But the #OMN second question is what happens when production grows faster than the ability to filter, discuss, trust and maintain? More books, but more noise, More apps, but more clutter. More papers, more pressure on review systems, more music, but harder to value human creativity.

The #dotcons logic says: more content = more value. The #openweb lesson is different – value comes from communities, trust, context and care. We don’t just need more production, we need better commons, better mediation and better ways to separate signal from noise.

The current wave of generative AI (#GenAI) is presented as inevitable, the message is everywhere: adapt, adopt, integrate, or be left behind. But technology is not neutral, as every tool carries assumptions – who benefits, who controls, what values are embedded, and what damage is accepted as “the price of progress”.

From a #OMN perspective, the question is not simply “can this technology do impressive things?” Of course, it can. The question is what kind of society does this technology build? Does it strengthen human creativity, collective intelligence and open participation? Or does it deepen the existing #dotcons path of centralisation, extraction, dependency and enclosure? The promise and the reality of large language models (#LLM) represent a technical development, they can summarise information, translate languages, generate text, assist coding, and help people interact with large amounts of information. These are real, if floored capabilities.

But the current #techshit hype jumps from useful assistance to much bigger claims: that these systems will replace expertise, solve social problems, revolutionise education, transform science, and create a better future. This is currently not true, and, on the LLM path will never be true as the current GenAI systems do not understand the world. They generate likely patterns based on huge amounts of training data. They do not know truth from falsehood, meaning from appearance, or ethics from probability, a convincing answer is not the same as a system that understands. This matters because the native #openweb was built on a different idea, that knowledge comes from people, communities, discussion, correction and shared responsibility.

The #geekproblem is confusing capability with wisdom is a recurring problem in technology culture – it is the assumption that if something can be built, it should be built. The technical question becomes “Can we?” while the social question “Should we?” gets pushed aside. This is part of what #OMN calls the #geekproblem – the tendency to reduce complex social questions into technical problems. A better search algorithm does not automatically create a healthier information system, a faster way to generate content does not automatically create better knowledge. More automation does not automatically create more freedom. The missing piece is the social context around the technology.

Then we come to the ecological cost of scaling, the current GenAI boom depends on enormous infrastructure. In the era of out of control #climatechaos data centres require huge amounts of electricity, water for cooling, specialised hardware, constant replacement cycles leading to massive extraction of resources. At a time of #climatechaos, we should be asking whether increasing consumption is the only path available.

The lesson is not that technology is bad, the lesson is that technology without social responsibility becomes a tool for whoever already has power. The question is not “how do we make AI bigger?” more it is how do we make technology serve human communities rather than making communities serve technology control systems, it is about who controls. The current dominant systems are owned by a few powerful companies controlled by the #nastyfew actively working to destroy our ecology and societies.

The future is not decided by whether we use AI, it is decided by whether we allow the same old #dotcons logic to shape every new technology. The work remains the same to build alternatives, keep processes open, grow the commons. The answer is not simply rejecting technology, the #openweb has never been anti-technology. The question is what kind of technology grows from what kind of culture. We need tools that strengthen human networks, not replace them. Tools that support commons, not enclosure, that increase agency, not dependency.

If we change this can there be an ethical AI? A socially useful technology? Possibly, but it would require a very different path, it would need many of the things the #openweb has argued for from the beginning.

#OMN #OGB #4opens #openweb #FOSS #indymediaback

Beyond AI

The biggest question is not whether #AI becomes useful. It is who shapes the surrounding paths? A future controlled by a few #dotcons will reproduce the same mess we have now of centralisation, extraction, enclosure. Were a future built through #4opens paths would look different.

The #geekproblem is believing the next tool solves the old problem. But many problems are not tool problems, they are relationship problems. The next stage is not replacing humans with smarter machines, it is building better human paths that can use machines without becoming dependent on them. Beyond AI is about making communities capable, the real upgrade is not artificial intelligence, it is collective intelligence.

AI is changing the scale of content creation, but not raising the quality. Generative AI tools have lowered the barrier to producing average books, apps, music, legal documents, academic papers and endless streams of text. The result is a massive increase in output, but what happens when production grows faster than our ability to filter, discuss, trust, maintain and give meaning to what is produced?

More books, but more noise, more apps, but more clutter, more papers, but more pressure on systems of review, more music, but a harder struggle to recognise human creativity and care. The #dotcons logic says – more content = more value – were the #openweb lesson is different, value comes from communities, trust, context and care. The challenge is not creating more things, the challenge is building better commons around the things we create.

The AI question is bigger than the technology, as the current wave of generative AI (#GenAI) is presented by our #fashionistas and there servants as inevitable. The message is everywhere to adapt, adopt, integrate, or be left behind. But technology is never neutral, every tool carries assumptions about who benefits, who controls it, what values it embeds and what damage is accepted as the “price of progress”.

From an #OMN perspective, the question is not simply “Can this technology do impressive things?” Of course, it can. The real question is “What kind of society does this technology build?” Does it strengthen human creativity, collective intelligence and open participation? Or does it deepen the existing #dotcons path of centralisation, extraction, dependency and enclosure? This is the wider #openweb question we should be focusing on.

Large language models (#LLM) and generative AI systems represent a real technical development. They can summarise information, translate languages, generate text, assist coding and help people interact with large amounts of information. These are useful capabilities, but the hype jumps from assistance to much larger claims – That AI will replace expertise – That it will solve social problems – That it will transform education and science – That it will create a better future automatically.

The problem is that current AI systems do not understand the world, they generate patterns based on huge amounts of training data. They do not know truth from falsehood, meaning from appearance, or ethics from probability. A convincing answer is not the same thing as understanding.

The missing social layer in our narrow conversations is that the #openweb was built around a different idea, that knowledge comes from people, from communities, discussion, correction, disagreement and shared responsibility. This is where the #geekproblem appears – the tendency to confuse technical capability with social wisdom – the technical question becomes “Can we build it?” the social question “Should we?” often disappears.

A better search algorithm does not automatically create a healthier information system, a faster way to generate content does not automatically create better knowledge. More automation does not automatically create more freedom. The missing piece is the culture around the technology, as technology without social responsibility becomes a tool for whoever already has power.

This is not even touching on that the ecological cost of scale is a catastrophe in the era of #climatechaos and social backdown. The current AI boom depends on enormous infrastructure, huge amounts of electricity, water for cooling, specialised hardware with constant replacement cycles leading to the large-scale resource extraction. At a time of #climatechaos, we should question whether endless expansion is the only possible future. The #dotcons model has always worked through scale, more users, more data, more infrastructure and more dependency. Generative AI is arriving inside the same economic system that created the catastrophic problems it claims to solve.

Then we have the open internet problem, the #openweb was built around participation, people created #4opens websites, communities, documentation, software and culture. GenAI introduces a different path, that the internet becomes raw material, this human creativity becomes training data. Communities produce knowledge, while large companies extract and monetise it. This creates a dangerous cycle were there is less support for creators → less motivation to create → less genuine knowledge → more dependence on generated content. Its #KISS to understand that healthy commons cannot survive if everything is extracted and nothing is returned.

The #Fediverse and the question of growth, a few years ago there was a feeling that the #Fediverse development culture was running on leftovers. Social movements arrived in waves, and many feared that more waves was moving into #mainstreaming. Since then, the Fediverse has grown, with more people knowing about decentralised social media, more organisations paying attention. Ideas that once lived mostly in activist and technical circles have moved closer to wider adoption.

But growth always creates a question – What happens when a movement becomes successful enough that the surrounding culture starts changing it? The early #openweb was built around different assumptions – People have agency – Communities shape their own spaces – Experimentation matters more than optimisation – Trust matters more than control and Commons matter more than platforms. #Mainstreaming brings pressures, these are not automatically bad. But there is a danger that the technology scales while the culture that created it gets diluted. Federation is a technical idea. Living commons is a social one, the challenge remains – now do we grow without losing the roots?

The narrow lesson from #FOSS – it is one of the greatest successes of the #openweb era. Without it there would be no Linux, no Apache, no Firefox, no Wikipedia-scale infrastructure and no Fediverse ecosystem as we know it. It has created extraordinary shared value, but success should not stop us asking difficult questions. The question is not whether FOSS works, the question is – Who does it work for? Where does it struggle? What social lessons can we learn? One recurring problem is the idea that open source is simply a marketplace of independent individuals.

When building the future we actually want – The question is not whether we use AI, more It’s whether we allow the same old #dotcons logic to shape every new technology. The future depends on whether tools strengthen human networks or replace them. Whether they support commons or enclosure, whether they increase agency or dependency.

But what we are seeing is that the tools we need most are often the first things stressed, messy and elitist systems try to defund, discredit and dismantle. Why? Because they require uncertainty, require questioning assumptions, require admitting complexity. Those are not weaknesses, they are survival tools.

Keep this in mind on native #openweb paths.

There is no intelligence in AI – and no path to any

Despite the constant #mainstreaming hype, the branding, and the trillions of dollars being poured into it, there is a simple reality that needs to be stated plainly: There is no intelligence in current “AI”, and there is no working path from today’s Large Language Models (#LLM) and Machine Learning (#ML) systems to anything resembling real, general intelligence.

What we are living through is not an intelligence revolution, it is a bubble – one we’ve seen many times before. The problem with this recurring mess is social, as a functioning democracy depends on the free flow of information. At its core, democracy is an information system, shared agreement that knowledge flows outward, to inform debate, shape collective decisions, and enable dissent. The wisdom of the many is meant to constrain the power of the few.

Over recent decades, we have done the opposite. We built ever more legal and digital locks to consolidate power in the hands of gatekeepers. Academic research, public data, scientific knowledge, and cultural memory have been locked behind paywalls and proprietary #dotcons platforms. The raw materials of our shared understanding, often created with public funding, have been enclosed, monetised, and sold back to the public for profit.

Now comes the next inversion. Under the banner of so-called #AI “training”, that same locked up knowledge has been handed wholesale to machines owned by a small number of corporations. These firms harvest, recombine, and extract value from it, while returning nothing to the commons. This is not a path to liberal “innovation”. It is the construction of anti-democratic, authoritarian power – and we do need to say this plainly.

A democracy that defers its knowledge to privately controlled algorithms becomes a spectator to its own already shaky governance. Knowledge is a public good, or democracy fails even harder than it already is.

Instead of knowledge flowing to the people, it flows upward into opaque black boxes. These closed custodians decide what is visible, what is profitable, and increasingly, what is treated as “truth”. This enclosure stacks neatly on top of twenty years of #dotcons social-control technologies, adding yet more layers of #techshit that we now need to compost.

Like the #dotcons before it, this was never really about copyright or efficiency. It is about whether knowledge is governed by openness or corporate capture, and therefore who knowledge is for. Knowledge is a #KISS prerequisite for any democratic path. A society cannot meaningfully debate science, policy, or justice if information is hidden behind paywalls and filtered through proprietary systems.

If we allow AI corporations to profit from mass appropriation of public knowledge while claiming immunity from accountability, we are choosing a future where access to understanding is governed by corporate power rather than democratic values.

How we treat knowledge – who can access it, who can build on it, and who is punished for sharing it – has become a direct test of our democratic commitments. We should be honest about what our current choices say about us in this ongoing mess.

The uncomfortable technical truth is this: general #AI is not going to emerge from current #LLM and ML systems – regardless of scale, compute, or investment. This has serious consequences. There is no coming step-change toward the “innovation” promised to investors, politicians, and corporate strategists, now or in any foreseeable future. The economic bubble beneath the hype matters because AI is currently propping up a fragile, fantasy economic reality. The return-on-investment investors are desperate for simply is not there.

So-called “AI agents”, beyond trivial and tightly constrained tasks, will default to being just more #dotcons tools of algorithmic control. Beyond that, thanks to the #geekproblem, they represent an escalating security nightmare, one in which attackers will always have the advantage over defenders, this #mainstreaming arms race will be endless and structurally unwinnable.

Yes, current #LLM systems do have useful applications, but they are narrow, specific, and limited. They do not justify the scale of capital being burned. There are no general-purpose deliverables coming to support the hype. At some point, the bubble will end – by explosion, implosion, or slow deflation.

What we can already predict, especially in the era of #climatechaos, is the lost opportunity cost. Vast financial, human, and institutional resources are being misallocated. When this collapses, the tech sector will be even more misshapen, and history suggests it will not be kind to workers, let alone the environment. This is the same old #deathcult pattern: speculation, enclosure, damage, and denial.

This moment is not about being “pro” or “anti” technology. It is about recognising that intelligence is social, contextual, embodied, and collective – and that no amount of #geekproblem statistical pattern-matching can replace that. It is about understanding that democracy cannot survive when knowledge is enclosed and mediated by #dotcons corporate capture beyond meaningful public control.

To recap: There is no intelligence in current #AI. There is no path to real AI from here. Pretending otherwise is not innovation – it is denial, producing yet more #techshit that we will eventually have to compost. Any sophist that argue otherwise need to be sacked if they arnt doing anything practical.

The only question is whether we use this moment to rebuild knowledge as a public good – or allow one more enclosure to harden around us. History – if it continues – will not be neutral about the answer.

What Did We Learn from Web3, Crypto?

Looking back from the mid-2020s, the arc of #web03, #NFTs, and blockchain culture is very clear. What once promised (lied about) decentralisation, liberation, and a break from corporate capture now looks like the same, mess, #techcurn pattern repeating itself, yes it had new language, new branding, but it was easy to see it had the same underlying mess making dynamics.

As these #geekproblem projects hollowed out, the signs became hard to ignore. The technical optimism faded, the user bases thinned, and the economic logic exposed itself. What followed was totally predictable: spin. Makeup and perfume slapped onto decaying projects to hide the smell of rot and exploitation. Rebrands. New narratives. New demographics. Same extraction. This was the outcome of building “liberation tech” on foundations that still centred virtical money, speculation, and power concentration.

With these projects we are now in the zombie phase, projects kept moving, kept talking, kept selling – long after the animating ideas had died. Influencers and promoters continued to perform belief, even as any substance drained away.

This is a few years when #fashionista culture met #encryptionist ideology – aesthetics and technical absolutism snogging the undead remnants of a failed #deatcult vision. The result wasn’t in any way decentralisation; it was a simply a new enclosure. People weren’t being freed, they were being financialised, the money problem #KISS

At the core was a simple structural truth: #dotcons feed on money. Put money in, influence comes out. That logic doesn’t disappear just because you wrap it in cryptography or decentralised rhetoric. “Bad actors” weren’t anomalies – they were following the incentives as designed. Aany social good becomes just collateral damage. This is why the lie collapsed in te end.

The deeper harm and problem with #techcurn is each wave claims to have fixed the problems of the last. But each wave reproduces them, because this is what works when worshiping a #deathcult. This isn’t just a failed tech trend, the #techcurn disparity, driven by extraction systems cause enormous human harm, displacing livelihoods, concentrating power, and amplifying inequality at planetary scale.

These systems don’t fail harmlessly, they fail onto people. That’s why the call isn’t just to “be critical,” but to step away – and help others step away too. Not through purity exits or individual moralising, but through collective paths back to technologies built for people rather than profit, life over zombies

There has always been another path: the #openweb. Messy, imperfect, slower, less glamorous, but grounded in shared infrastructure, social trust, and human-scale governance. The #OMN approach doesn’t promise salvation. It offers compost instead of speculation. Process instead of hype. People over tokens.

A note on hashtags: And yes, the hashtags matter. Click them., search for them. They cut sideways through algorithms – small back doors into less mediated, less controlled ways of seeing. Not a solution, but a crack in the wall.

The current #Ai hype bubble is repeating this mess with a little more useful #LLM functionality, but on top of this is a huge mess of #techchurn, which will need composting.

Observation: some people go into news to speak truth from power – using institutions to legitimise the status quo and defend the worship of the #deathcult.

Others speak truth to power – using journalism to expose, question, and challenge unequal power and its consequences.

Only one of those serves the public interest #KISS

Trust, encryption, and the risk of “trustless” thinking

Telegram messaging app is dieing

Telegram partnering with Elon’s #AI to distribute #Grok inside chats is a clear line crossed. This matters because private data ≠ training fodder, bringing Grok (or any #LLM) into messaging apps opens the door to pervasive data harvesting and normalization of surveillance.

This is an example of platform drift: Telegram was always sketchy (proprietary, central control, opaque funding), but this is active betrayal of its user base, especially those in repressive regions who relied on it.

Any #LLM like Grok in chats = always-on observer: Even if “optional,” it becomes a trojan horse for ambient monitoring and a normalization vector for AI-injected communication.

“Would be better if we had not spent 20 years building our lives and societies around them first.”

That’s the #openweb lesson in a sentence, that the #dotcons will kill themselves. This is what we mean by “use and abuse” of these platforms which have been driven by centralization, adtech, and data extraction, that they inevitably destroy the trust that made them popular. It’s entropy baked into their #DNA. As Doctorow calls this #enshitification, the tragedy is how much time, emotion, and culture we invested in them – only to have to scramble for alternatives once they inevitably betray us.

What to do now, first step, remove data from your account then delete telegram app, not just for principle, but for your own safety. Move to alternatives – #Signal for encrypted, centralized messaging (trusted but closed server). There are other more #geekproblem options in the #FOSS world but like #XMPP, #RetroShare, or good old email+GPG can work too, but they can be isolating, so stick to #signal if you’re at all #mainstreaming.

Then the second step, build parallel #4opens paths by supporting and develop alt infrastructure like the #Fediverse (Mastodon, Lemmy, etc.), #OMN (Open Media Network – decentralized media), XMPP and #p2p-first protocols, #DAT/#Hypercore, #IPFS, or #Nostr etc.

Yeah, things will get worse before they get better, what we’re seeing now is the terminal phase of the #dotcons era. These companies are devouring themselves and will eventually collapse under the weight of their contradictions. The question is, will we have built anything to replace them?

If not, authoritarian tech (like Elon’s empire) fills the void. That’s why we rebuild the “native” #openweb, even if it’s slow, messy, and underground. That’s why projects like #OMN and #Fediverse matter. If you’re reading this, you’re early to the rebuild, welcome, let’s do better this time.