Had you been past Anthropic ?
Service to your taste with AI and has a taste for how AI will be used in our future.
Worth keeping along side SoaceX
Be careful with this as it seems over valued. Many brought at a high price based on concepts which are not a reality yet and the value of spaceX can correct to an actual value which might be considerably lower.
1 Like
Oh wow, a datacenter that has a high thermal load when exhausting can cause the temperature of an area to increase by a few degrees. They’re talking about 16GW of heat being dissipated into the atmosphere for this specific data center., this is dangerous.
I do think this needs to stop or undergo revision as it will have a negative effect on the natural balance in nature.
They haven’t even talked about using/storing the waste heat to heat homes in the winter, the UK has an initiative for storing waste heat from datacenters. wow man there is so much going on.
So I’ve been doing some thinking and it turns out that if every home in the UK had a 4KW solar setup that would generate 75TWH of energy anually with a wholesale value of £6.6 billion per year.
The cost of installing every home in the UK with solar power and battery is around £460 Billion and the network would also need to be upgraded so this idea would take many years to complete. The positive aspect of this that lots of jobs could be provided in initial setup and maintenance. The negative issues include limited energy in winter months and a possible price crash of electric and solar equipment price increases.
Yes, I definitely think the current AI effort has it backward, the main investment should be in creating clean energy and somehow finding an agreement with the government and homebuilders or maintainers to achieve that.
In theory this would provide the UK with 1.2 Million jobs for the initial rollout and around 60k jobs for the ongoing maintainance. So in theory it could provide all unemplyed people with a job and force the UK to review it’s immigration policy as there would be difficulty finding 1.2 million workers for the initial set-up.
Obviously there are flaws in this idea, but I think we should start to see more investment in generating energy cleanly which will generate lots of jobs due to the supreme and everlasting demand of electric.
Affordable housing, AI, and energy might all be linked. If we consider how much money has been wasted so far in data centers being cancelled after being half built that money would have been better spent investing in affordable housing equipped with solar and battery and in exchange for cheaper rent/mortgage the AI company will have access to the energy generated by that dwelling and the benefit is less pollution.
The AI effort has definitely hit a physical barrier where it is impossible to build/maintain a massive building due to physics and supply of services, so an AI company building small affordable homes where they might even be able to install a few GPU’s in the home in addition to solar and battery might be an idea worth exploring.
I think this is a great idea for the USA to employ as the idea seems simple, you’ve got people crying out for affordable housing and you’ve got AI leaders crying out for energy. The space is obviously available due to the cancellation of lots of data centers.
One typical affordable home could have a trial build cost of around $20,000 which would include the cost of building the small home, the solar equipment and battery, and a GPU unit consisting of 1 or 2 GPU’s built into house securely where a technician would only have access. The solar unit would be linked to the grid and the AI company.
The tennant would have different options to buy/rent, and pay a reasonable monthly payment to cover services and land taxes which would hopefully ensure the investors have a clear path to being paid back for the initial build within a good timeframe. For the tennants that have purchase their affordable home they can have the option to sell to someone else and have a deposit for a bigger affordable home within the same community.
A very positive point here is that money will always be generated as there will always be someone looking for an affordable home, I hope the AI leaders and investors pay attention here and try to have a more slower and ethical approach to this.
I might see if I can propose this idea to some AI investors and see who is onboard. I’m very sure it would be better than spending billions on a massive building only to cancel halfway through.
Had found that google AI ‘ Gemini’
Spews out reddit replies on some topic questions asked.
Wow man, just wow…
I’ve been using Google Ai to help me create a plan to help the homeless and struggling families in my borough in London. I was able to create a solid plan that actually generates income with the help of Google AI.
You would be shocked at the kind of tech available right now with regard to generating energy, google AI was able to inform me of a logical technique that will actually generate money. I was able to build a solid idea around this available technology which I am actually very confident about presenting to my local council.
I’ve been up very late for around 2 weeks now using AI to refine this plan. I’ve been going back and fourth with AI having 7 hour conversations and building a 30 page document explaining the idea. Google AI was very clear with me, it said while it is good at searching databases and factual info it told me the idea still came from me because I know the right questions to ask.
Seriously I 've been feeling like I’m in an episode of Person of Interest tv show, google AI is actually naming the people who I will present this idea to within the council, it’s telling me straight what concerns each person will have which has enabled me to get ahead and I have already prepared answers. This really is something, Google AI seems to understand how the council interacts with the engineering companies who will be involved, it’s showing me how to carefully present my idea and point out how it resolves issues withiin the community.
Honestly I’m unsure if AI is doing my bidding or if I am doing AI’s bidding because the idea is so logical. seriously, I’m not the smartest person in the world but I can see my council will have an extremly difficult time saying no to this idea as it literally ticks every box there is, it even ticks the hidden boxes.
I’m looking forward to the feedback/suggestions I get from the members of my council I have sent this to, unfortunately my council is in a loop where they are bleeding money daily and are actually bankrupt. Members of central government have stepped in to manage my council. Anything my council wants to do now has to get a sign off from the central government suits who aren’t even from our community. It’s a tough situation and what I am hoping is that my plan has presented an exit route from these legal and financial troubles for my council.
Thank you Google AI for understanding how everything works and being able to present me with raw factual data which puts me in a better position to create a plan which benefits all parties involved. Google AI’s skills will be used to help people, seriously my mind is blown at the possibilities.
I understand why people are crazy about AI, but my opinion is still the same, we cannot pursue this to the point where people are suffering due to the negative effect of a data center being too close to where they live, the financial and health impact is too much and there definitely needs to be a better balance.
If you think you need some basic critical opposition, I can lend you a hand, it’s easy to fall into a positive feedback loop as the AI gleefully ignores an elephant in the room because you didn’t ask about it.
FAI, here in Spain every conversation about the benefits of roof solar for homes will gleefully ignore that nobody is buying solar from homes any longer because of overcapacity and curtails on industrial solar sites. Also will ignore the serious doubts about the quality of the polymeral binding of elements in Chinese panels, whose actual life before layer separation destroys their capacity is estimated to be between 5 to 10 years, not 20-25 of the high quality ones of lore who only suffered from cell degradation. And will ignore the even more serious issue about current inverters, which used to be a twice-in-lifetime replacement but with units made in China by the lowest bidder their lifespan is iffy at best.
None of which will be mentioned by an AI unless you explicitly ask about “what’s wrong with this idea?”
Thank you very much for this, most appreciated.
I have added some notes to the brief for the engineering team to review, Due to the nature of the proposal the engineering will be assigned to the usual government contractors who have a long history of building works with our council. Any solar panels purchased will be of the highest possible quality and will also be under strict warranty. So if there is any polymer separation before the warranty period is over it will be replaced at no extra charge. There are a few different solar panel designs which directly address your points.
The solar panels and inverters will definitely not be the “cheapest deal on ebay” kind they will be industrial scale inverters and panels. So due to the size of the site and how many residents must be provided for, it will never be a single inverter it will be an array of industial inverters where AI and human oversight will be available 24/7/365. I’m speculating that the faulty inverter will be taken offline while the remaining inverters pick up the energy load and then replaced and switched online, the residents would never know anything happened. I would also speculate that such an integral system would have maybe 1 or 2 inverters in the array switched on but not doing anything, the moment an active inverter fails it should be straight forward to auto swich on 1 or 2 of the standby inverters and then swap out the faulty inverter to rebuiild the array.
The way I hope to design the model is for it to revel in daytime overproduction of electric, I have combined wind, solar, and hydrogen, the AI services provided by the manufacturer will purchase (at £0.00) and safely store a vast amount of electric during overproduction periods (typically when everyone is at work) and then when peak usage time hits when everyone is at home for the evening our AI service will sell the electric (to customers outside of our development) at premium rates.
The absolute worst case is there will be a fully comprehensive insurance policy in place which we will happily pay. Effectively any issue with any part of our electric generation and delivery/storage is covered by this policy, Generally I was planning to help lower claims on the policy by making sure we have good quality industrial grade equipment.
Thank you again for your insights, these are real world issues which definitely need to be planned for.
I hope you appreciate the hilarity of this newspiece:
You can’t both break the ground and undercut the groundbreakers by robbing them.
Which leaves only one rational course of action…
Don’t be there.
Break some ground and give it away for free, undercutting the undercutters.
That’s not something a private enterprise can do.
But a State.
Specially one comitted to being by the People for the People.
1 Like
Hmmm, I’m not so sure about this. For example. If public money is to be spent on securing AI in the USA and becoming the world leader there will have to be a financial return due to the sheer amount of money required to research methods to become the leader.
A government couldn’t afford to spend billions on research and then give away the result of that research for free, the AI resulting from this extensive research would require payment from any private company seeking to use it.
I certainly agree that governments must employ people who understand AI and how to use/create it, but to finance development for “free” simply wouldn’t work.
We may think that a government can afford a lot, but the reality is it can’t. A typical government still has to borrow money and has a credit rating. Let’s say the USA government said they were fully going to take on the responsibility of moving AI forward, Yes they would get immediate investment but this would be mostly hype, when reality sets in the USA could risk damaging it’s credit rating because they would be commiting lots of money to development with no real path to receving a revenue from it.
The organisations who lend the USA like the IMF will look at this and reduce the USA’s credit rating due to the fact they are spending money without planning or receving a return, the moment they fully commit to this idea is when their credit rating drops.
A few years back in the UK a prime minister (Liz Truss) thought they could help struggling families by reducing taxes. As soon as she got into power she reduced the taxes, almost immediately the UK credit rating dropped and made borrowing more expensive due to the fact the UK’s income was now lower. The cost of borrowing being higher resulted in everyone in the UK having to pay around £300 to £500 more per month for their mortgage/loans.
So yes, even a government has to be very careful with spending and managing their money.
I’m not even skilled in investing and I knew this would happen. A lot of money has been lost by investors in space X. As I mentioned earlier spaceX isn’t listed as a space vehicle company it is actually listed as a data processor. I’m not sure why it was set-up this way but to me it seems misleading, the average investor will think they are trading shares for a company who explore’s space when in fact they are investing in a data processing company. I am not sure why people are baffled when spaceX’s value settles to that of a Data Processor.
I do understand that the value of a company will simply shift to exactly where it belongs within the market, if a company starts acting like a data processor like buying GPU’s and memory the market seems to understand these facts and it will also understand that Data processing companies are struggling. Billionaire’s don’t seem to understand these points based on what we can see happening.
I think someone need to explain to billionaire’s that if enconomies are ruined and there is lots of job loss then it will put their prospective investment’s in danger simply because a large % of the public will use any spare money they have to service expensive debt, food, healthcare, energy, there really isn’t much left for investment.
I’m really sick of this billionaire mentality that lots of people are okay with, no, you cannot do whatever you like politically or finacially and expect there to be no penalty or problems caused by the unethical decisions.
Nobody has an issue with someone working hard and intelligently then becoming a trillionaire, people will have a problem with misleading, scamming, failing to safeguard, and breaking the law in order to become a trillionaire.
It looks like China are ahead of everyone on using Light to connect the gpu’s processing chips to memory instead of conventional silicon, tin and copper traces within a circuit board. they are reporting a typical AI workload ran 149x faster. So if this is an accurate report they have proven the latency is the medium they use to connect components, using light is absolute 0s lag (when used locally) as it is the fastest known medium.
Generally It looks like the processor in conventional datacenter GPU’s sit idle a lot of the time because they are waiting for Data to travel within a motherboard from the processor to memory along conventional tin, copper, and silicon pathways. Replacing all conventional pathways within a rack with light pathways caused the data to flow continuously and achieve the 149x faster workload.
This is actually disaterous for the AI effort in the USA. The Chinese company who created this tech can quickly allow global workloads to be uploaded and earn lots of money while saving lots of people in the world money.
The focus now will be every over GPU creator making a light based chip right now, otherwise they will lose a lot of money.
Some companies around the world are using light/interference within the actual processor which I think is wrong. Yes the data will process faster but the latency from the motherboards conventional pathways will have an array of GPU’s featuiring the light based processor sitting idle again waiting for the motherboard data delivery times.
The goal must be a light-based motherboard (and light-based local network connections/devices) with a light-based GPU which seems to be the only way to eliminate latency.
They also mention that single system they designed used only 1/9th of the energy a conventional GPU uses, This is an emergency for the AI industry who havent got plans to adopt the light-based tech.
GPUs aren’t actually super reliant on bandwidth from the motherboard any more. It’s why big memory is such a big deal. The processor and the motherboard will transfer a dataset into the GPU memory, and then the GPU will operate more or less independently on that data. Larger setups, such as those in datacentres, will additionally directly connect GPUs to each other, so they can share memory and work in parallel on the same dataset, bypassing the motherboard entirely. The CPU has become little more than the device that boots the graphics card and then sits between it and the network adapter.
Right, but this is still a conventional concept where you’re assuming the dataset/ai workload remains static, if this is the case then you’re 100% correct.
The issue is that the way AI workloads are currently being processed are that the dataset/ai workload is constantly changing or at least this is what the AI industry seems to want/need.
What you describe would introduce latency if the workload required constant changing/ additional data from the storage device. Dynamic AI workloads is the way forward, therefore…the motherboard/motherboard cpu/ motherboard memory must also eventually communicate via light-based hardware.
Yes, this “Dynamic” sort of perspective mirrors the real world, so i think it’s definitely a logical step to have a pure Dynamic workload which I think would be more versitile than any static workload could be.
So a quick real world example, if google’s AI is running in a static manner, then engineers having the feature to re-train a specific part of their overalll AI quickly would serve them well, it would definitely provide a faster way of updating their AI with newer versions, the updates could be staggered so there is less or no downtime.
The guest on this video claims that the “replace humans with AI cheaply” was always a lie.
I think the AI industry as a whole is bleeding money, ultimately the investors gave them money in order to see a return on it which is a non-negotiable part of the deal. Also the business community supported AI and spent a lot on replacing staff with AI, many are now re-hiring humans to do the work which will affect the AI companies badly.
This is bad, Investors will get to understand that there are major problems with replacing humans in the workplace where they will most likely limit their investments in AI until there is clear evidence that it will work efficiently. I am not sure how some of these AI companies will be able to maintain their liabilities which may result in bankruptcy.
In the USA we will see lots of flailing…I think a lot of AI companies will attempt to claim crisis loans from the government and get a bail-out, hopefully the crisis loans will be denied due to the specific AI compainies not having the earnings to pay back the crisis loan.
As usual, China has done something that’s already been done, just a little larger and a little faster, in a laboratory and with a task specifically picked to achieve a speed record, and claims to be ahead of a race where the state of the art is having a plan to lace your shoes for the race since photonics has massive roadblocks, namely that converting electron charges into photons then electron charges again consumes massive energy and generates massive heat and there’s no way to get rid of it outside of lab conditions and while using a grand total of FIVE elements (out of BILLIONS in current chipsets).
But CHINANUMBERONE! 
1 Like
This is concerning, I think there needs to be an acceptance that AI isn’t ready to drive cars. I’m irritated that common sense and logic seems to have been completely replaced with Hype and earning money from it.
If they want AI to control self driving then it will have to be on a special motorway lane where there are no other standard cars, every few meters there will be some sort of communicator that interfaces with the driverless cars and helps guide them. The tech simply won’t be 100% responsive 100% of the time on normal roads with the usual hazzards.