Written By:
- Date published:
11:50 am, June 21st, 2026 - 21 comments
Categories: Deep stuff, Economy, Politics, tech industry -
Tags:
AI is fascinating and I use it every day. But the more I use it and the more I think about it the more issues I see with it.
Like these.
Environmental effects
All these data centres that suck up huge amounts of energy and water are not good for the environment. Many local communities are already engaged in fights to oppose the creeping incursion of data centres into their neighbourhoods.
The data centres draw huge amounts of power and water.
In the United States it was estimated that in 2024 data centres consumed 4.4% of the country’s total power generation. And that by 2028 this figure could be as high as 12%. At a time when the world is struggling to reduce greenhouse gas emissions this is a whole lot of extra CO2 generation.
As for water estimates are that next year data centres will consume up to 6.6 billion cubic metres of water, half of the United Kingdom’s total consumption.
And get this. In the next 12 months it is estimated that the world will spend up to $1.37 trillion US on AI infrastructure.
Compare this to what other areas the money could be spent on.
We could solve world hunger for $265 billion a year.
This spend could also go a long way to resolving the climate crisis. It is estimated that it will cost the world $3.5 trillion per year to develop low emissions assets. Every dollar spent on a power hungry data centre is one dollar less that can be spent on a solar panel.
Concentration of wealth
The big tech companies engaged in AI are all owned by wealthy people. And their motivations are not altruistic. They are doing this to increase their wealth and power, not for the common good.
You just have to look at the major investors in AI. Mictosoft, Google, Jeff Bezos’ Amazon, Mark Zuckerberg’s Meta, and Larry Elison’s Oracle are involved heavily. These people display advanced levels of cleptocracy and are not motivated by a desire to contribute to the common good.
Remember the Billionaire Space Race? This feels like similar testosterone driven big dick energy being engaged in by the world’s most powerful men while the world is being held hostage.
And check out who they make political donations to? Concentrating the wealth in the hands of people so willing to sacrifice the political system and the common good so that they become even wealthier is not good for our world.
Disruption to society
AI is already affecting employment trends and reducing job opportunities.
It has distorted hiring activities. There is already a new generation of graduates who do not have jobs because businesses think that AI can do the job again.
And there is no sign of a collective response to the issues. And Big Tech is making larger and larger donations to political candidates that pose no threat while at the same time seek to undermine those who do.
AI is evolving rapidly. And collectively we still have no idea what we will be doing with it.
Disruption to the economy
We are in the bubbles of all bubbles. Apart from Nvidia which sell the chips that AI depends on we have a host of AI companies making huge losses and showing no sign of becoming profitable.
They are burning through cash. The current business model is to take on new clients but this makes their losses bigger because the cost of providing the service is greater than the income derived from providing the service and is not predicted to change any time soon.
Yet despite this their valuations are sky high. Many commentators think that we are in the middle of an almighty bubble.
And with SpaceX just listed on the Nasdaq and Anthropic and OpenIA planning to do the same thanks to recent changes in Nasdaq’s rules be very afraid. Their size means that your Kiwisaver will very probably find its way there through exchange traded funds. Musk is going to get some of your money.
As they say when the music stops it will be interesting to see who misses out. It could be a contest between Elon Musk and Larry Ellison.
And there are some really whacky ideas.
Like orbital AI data centres. Imagine what you would have to do to replace the most simple of computer part replacements?
If there is a correction the ripples will be massive. It could make the Dot Com crash in early 2000 look like a walk in the park.
Threats of an AI takeover
Remember this?
It may be not so dramatic. It may be that an AI process requested to perfect the making of paper clips gets carried away.
And there have already been instances indicating nervousness among governments.
Anthropic’s Fable 5 was recently removed from public access after security concerns were expressed by the US Government and it imposed export control directives. The worry was that the software could identify software vulnerabilities and help hackers.
The Government’s actions may have been related to Anthropic’s refusal to allow the Government to use its software for domestic surveillance and fully autonomous weapons systems.
Its reticence was wecome. In a Kings College study involving multiple war simulations AI decided to start throwing nukes in 95% of simulations.
AI should never be allowed near any military system decision making.
A possible response
These are privately owned assets that has been generated by strip mining the internet and the collective intelligence developed by humanity. We all have a collective interest in how this develops.
Bernie Sanders has been working on a response.
About AI he says this:
“Frankly, I think you have got to slow this process down,” he said. “It’s not good enough for the oligarchs to tell us, it’s coming, you adapt. What are they talking about? They going to guarantee health care to all people?
“What are they going to do when people have no jobs? What are they going to do, make housing free? So I think we need to take a deep breath, and I think we need to slow this thing down.”
His solution is to have a one time 50% tax on the stock of the large AI companies and to then place these stocks into a public fund. The Government would then hold voting rights and board seats and work to prevent harmful anti human decisions. Profits could be distribted to everyone.
Conclusion
The downsides are evident. Increased environmental damage, further concentration of wealth in the control of the richest cleptocrats on the planet, disruption to the labour market and to young people’s futures, potential devasatation to the economy if and when the bubble bursts and giving a program with control of military responses have an inclination to throw nukes around, this needs a considered and careful response.
But there is no sign of authorities National or International actually taking control of the situation and seeking to manage it. Instead we are trapped in this laissez faire response which is willing to trust the world’s future to the greedy inclinations of some pretty weird people.
Be afraid …
Fear as operational stance tends to make folks defensive. Hunker down. A proactive response would go with the flow. Disruptor tech is a primal directive for darwinism.
Humans evolved in a dynamic balance of order & chaos: nature features both continually. Order usually prevails but the theory of punctuated equilibria re-arranged academic thinking on that. Change emanates from the margins (chaos theory).
AI in this view is merely the latest trend via which uncertainty leverages mass expectations of the future. Being fearful can be adaptive, yet riding a tech wave is likewise – so surfers will always jump onto it.
Fear, or caution, is a survival instinct.
We often get told that something is the future.
Plastics, nuclear power, individual and ICE transport…
Just because huge industries profit from selling the shiny hype doesn't mean as a species we benefit. Cui bono with AI?
The good news is that global governance was proactive: https://www.japan.go.jp/kizuna/2024/02/hiroshima_ai_process.html
The problem with international law is lack of formal process, so this entrepreneural development led by Japan influences mass expectations of behaviour by key players,
Caterpillar, chrysalis, butterfly.
Humankind is in the process of crystallising (or cocooning, if you think of us as moths).
Our exoskeleton has formed and the process of dissolution has begun ; we've sloughed our mythological beliefs and practices and our structured religious forms and find ourselves cast-adrift in an unfathomable universe, clinging to a rapidly-unravelling thread we've called science. AI is the solvent that will finish the process, at least for those who have not fashioned themselves an anchor to reality. If you've not worked on yours, get busy – the Great Limination is upon us 🙂
Create a personal mythology for yourself and hold on; it's going to be (is) a bumpy ride 🙂
Have a thought of adopting Aldous Huxley as one of your gurus. He had an interesting and educational youth and developed into wise man with humour who liked women. That's quite well balanced I think but he had vision problems so probably was drawn to the personality of the real man and woman, rather than attraction first by looks.
https://www.brainyquote.com/authors/aldous-huxley-quotes
https://www.imdb.com/name/nm0404717/ – Aldous Huxley (1894-1963)
https://marketoonist.com/2024/09/aitool.html
I get a tinge of pain from the problem in the cartoon. Note that the screws are not only different sizes, some what I call Slot, Phillips, positron I think,.. When we say pollies have a screw loose, very likely, here's a site exhibiting 30 heads.
When we get robots in the pollies' place we will find that their core thought function will be replaced by stealth at night ('New heads for old' will soar from street-vendors' mouths) and the business of developing new connectors will be bigger than crypto-currency.
Thanks greywarshark – I hadn't noticed the different screw head types.
Wonder how close AI is to ‘computing’ human diversity – some of us have at least one screw loose. And then there are cross-threaded screws, stripped screws, etc. etc. etc.
Why would AI even bother?
https://marketoonist.com/2026/03/human-in-the-loop.html
Disclaimer: in my day job I'm a data and AI lead in a critical infrastructure industry, so I've spent a lot of time over the last year working through AI strategy, governance, security, and guardrails.
I think it's important that before we start arguing about AI, we're clear about what AI actually is, and what it isn't.
Using Microsoft Copilot to summarise emails, Claude Code to help write software, or ChatGPT to draft reports is not the same thing as autonomous weapons systems, military decision-making, or fully autonomous agents acting without human oversight.
Those are very different technologies, with very different risk profiles.
Nor is AI somehow beyond the reach of regulation or governance. It is entirely possible to run AI systems within strict organisational controls, including sandboxed environments, restricted data access, human approval workflows, audit logging, and locally hosted models.
Many organisations choose not to do this because it adds cost and complexity, but the capability certainly exists.
That doesn't mean there are no legitimate concerns. Questions around energy consumption, labour market disruption, market concentration, privacy, surveillance, and military applications are all worth taking seriously.
But I worry that discussions about AI often lump everything together into a single category. The risks associated with a coding assistant helping an engineer write Python are not the same as the risks associated with an autonomous weapons platform.
It's also important to note that the general experience of AI adoption so far is that it is most useful and cost-effective when it's used to amplify and supplement human beings, not replace them. The most successful implementations I've seen tend to make people more productive, rather than remove people from the process altogether. Human judgement, oversight, accountability, and domain expertise remain essential.
My experience has been that AI is neither magic nor uncontrollable. Most of the work is actually governance, controls, security, data access, and deciding where it is and isn't appropriate to use.
And yes, the infrastructure can consume enormous amounts of power and water and should be subject to appropriate regulation. But that doesn't mean we should throw the baby out with the bathwater.
Likewise, while the AI companies themselves are undoubtedly powerful, concentration of power is not a new problem. We already regulate banks, utilities, telecommunications providers, and pharmaceutical companies. AI companies should be no different.
The technology is already here, and people are already using it. The question is not really whether AI should exist. The more useful question is where it creates value, where it creates risk, and what safeguards we should put around it.
If we're going to have a useful conversation about AI, we need to be specific about which technology we're talking about, what risks we're concerned about, and what governance mechanisms are available to address them.
Thanks Res Publica. That is a very useful summary of AI issues.
Thanks RP. I tried to summarise a couple of years of concern into a relatively brief post. I agree nuance makes it much more complicated. There are benefits, I enjoy them myself.
The potential concencration of power is more than I have ever seen however. And we don't seem to be doing anything about this.
That's absolutely a legitimate and serious concern, and one that nobody seems to have fully solved yet.
I suspect part of the answer looks something like GDPR or a set of internationally recognised rules around AI governance, sovereignty, transparency, and guardrails.
The challenge is that the technology is evolving faster than most legislative and regulatory systems can keep up with. By the time governments understand one generation of the technology well enough to regulate it, the next generation is already emerging.
"But that doesn't mean we should throw the baby out with the bathwater."
Even if we wanted to throw out the bathwater, we couldn't – it's too late for that.
So, the baby – tell us about it! What's so wonderful about AI-baby?
Taking the baby thought – they, we think, start out as tabula rasa and whose minds are affected and effected very largely, by deliberate or involuntary input, but whatever there is waste output which must be monitored for the health of the system.
My bet in the next 3 years is that the Chinese open AI systems together doing drudge work, with their in-built hardware advantages in robotics, electric cars, weapons systems, and open programming systems, will beat the US super-high-capital intensive systems that are only used for very narrow and expensive services…
… and unlike the Chinese versions, out of the US AI super-high proprietary systems only a very few companies will survive.
But
big AIthe one percent has covered their arse, socialise the losses by forcing public wealth to invest at the top, and then after the crash they'll graciously buy back at the bottom./
US President Donald Trump is planning to meet the bosses of some of the country's most notable artificial intelligence (AI) companies to discuss the government taking a financial stake in their future.
Speaking on Air Force One, Trump said the goal of the US government investing in AI companies was to "create almost a partnership with the American public".
[…]
Although the president did not name specific companies, the biggest companies in the US working on AI are Google, Microsoft, OpenAI, SpaceX and Anthropic – the latter two of which are expected to go public in the coming weeks.
https://www.bbc.com/news/articles/c98r8r7dz5no
Would be infotaining enough to trigger a formal protest by the yank ambassador, huh? My rw younger brother always cited Obama's investment in that green energy corp that went bankrupt (Solutra or something). Obviously if the thing is run by bureaucrats it will fail bigly, but a smart leftist govt would just contract it out via tendering or any deal that incentivised suitable performance. My take on public/private partnerships is that failure is normally produced by collective idiocy at the design stage (more than in implementation). Anyway full marks for T's naivety in promoting a partnership with the public. Habituated to being perpetually screwed by govt, it should send them into mental disneyland for the forseeable future…
Do you see the problem with this?
Taking control of the stock does not necessarily require taking on the liabilities.
Rory Stewart has been banging on about the national security issues because of the problem of depending on American technology which they can turn off on a whim – as Trump did with one of Claude's models. As AI becomes important for doing things efficiently, not having AI will mean a country loses competitiveness. We effectively become blackmailable by those with the technology.
Makes me think of an old NZ Listener 'Letters' quip about fallout from the A
NZUS rift.