Thad Scheer

Principal and Chief AI Officer at @Sphere_OI. Lots of artificial intelligence and some actual intelligence. #AIGuy #AI #LeanDesign

[Thad Scheer @ThadOfSphere]

A sober and rational analysis. Nothing to seriously disagree with. But I think punting this once-in-a-lifetime opportunity to rid Iran of the IRGC and its clerical leadership is a disappointment. The words, “we should have finished the job” will echo in history.

[Jared Howe @Jaredhowe]

Trump killed 50 of Iran's top leaders, did at least $1 trillion worth of damage to their military equipment and infrastructure, totally wiped out their naval fleet and air fleets, put their government in a state of endemic internal conflict, and left them totally incapable of meaningfully attacking their neighbors in the region any time soon. By the end of it, all Iran could do is rig the strait with mines and take pop shots at passing ships.

The loss of insurance coverage had more of an impact on the flow of marine traffic than Iran did.

I personally would have preferred the total destruction of the IRGC (which technically isn't off the table if they decide to FAFO) but it would have been difficult to accomplish without more civilian casualties -- something third worldist leftoids were already complaining (and campaigning) about.

In the meantime, consider the following:

No boots on the ground. No prolonged conflict. No permanent occupation. No military draft. No lasting impact on energy prices. No multi-trillion dollar boondoggle. None of the things anti-war retards with Israel tunnel vision like Dave Smith and Candace Owens predicted would happen.

In addition:

- Oil and gas prices are falling in time for summer and the midterms.
- "Free Palestine" third worldists like Graham Platner have less ammo on which to campaign for Congress.
- Russia is bringing in less revenue.
- The U.S. got a foot in the door to block China's Belt and Road projects.
- Trump has greater latitude to tighten sanctions on Russia without exacerbating supply shocks to energy markets.
- Europe and America have agreed to increase cooperation in providing for the national defense of Ukraine.
- Israel isn't a signatory to the MOU and isn't bound by its terms, leaving it free to independently defend itself from threats if necessary.
- America isn't creating a power vacuum for China and Russia to fill by absconding and surrendering its own influence over the region.

The people who say this is a "humiliating defeat" for Trump?

- Russia
- China
- Iran
- Democrats
- The leftist Drudge-led media establishment

All the worst people you know have joined arms and are pretending like Iran pulled one over on the Bad Orange Man in hopes that it will piss off Republicans and Israelis enough to get them to kill the deal themselves.

It's very transparent. Don't let your enemies control you with such a stupid and obvious Reflexive Control op.

[Thad Scheer @ThadOfSphere]

@digijordan Hallucinated medical scans. What could go wrong there?

[Thad Scheer @ThadOfSphere]

The market for tokens is going to CRASH.

We already see cracks in the wall. It'll take a couple of big names to announce they are shifting to self-hosted open-source and the dam will break. Datacenter buildouts will freeze in their tracks.

[Thad Scheer @ThadOfSphere]

I'll just say, the 14-point memorandum between the U.S. and Iran is a disappointment.

So much more could have been achieved. Now, the regime continues, massively empowered. Free-minded Iranians will be slaughtered. And they'll still get a nuclear weapon, because...they lie.

[Thad Scheer @ThadOfSphere]

@scion_x_ You're right, except...

Despite the high utility of LLMs, the costs are unsustainable. So-called "real intelligence" might [should] run on a fraction of the energy and learn efficiently in the real-world (no more backpropagation).

That's utility.

[Thad Scheer @ThadOfSphere]

@marty_kausas For developers, what measurable KPIs would rationalize that $3K/month expenditure?

[Thad Scheer @ThadOfSphere]

@anilkseth @StuartHameroff The question is, did any of the 50 accepted proposals represent Stuart's position/side? I think the concern is that a 'legitimate' academic selection process is being used as cover for canceling off-narrative ideas and voices.

I could be wrong, but it often seems that way.

[Thad Scheer @ThadOfSphere]

@bendee983 Success is measured in headcount reductions. Bosses want one 10x engineer on the payroll and a team of bots doing the work that used to occupy 5-10 people.

We see this in RPA/automation. Headcount reduction is the only KPI that people give a s**t about.

[Thad Scheer @ThadOfSphere]

This is not the end, this is the beginning.

Fully autonomous drones have killed human soldiers for the first time. No human in the kill chain.

This was inevitable and there's no going back. This is what war looks like from now on.
https://t.co/b00lDph13g

[Thad Scheer @ThadOfSphere]

This is the future, we all know it.

Self host or go broke.

[starmex @starmexxx]

AMD CEO LISA SU HELD A MINI PC ON STAGE THAT RUNS A 235B MODEL AND REPLACES YOUR $440/MONTH AI STACK

amd's ryzen ai max+ 395 is the first x86 chip that runs a 200 billion parameter model on one piece of silicon. cpu and gpu share 128gb of unified memory, no separate graphics card needed

the gmktec evo-x2 runs qwen3 235b fully, deepseek v3 comfortably and llama 3.3 70b with headroom. on linux you get 110gb of usable vram out of 128gb

amd claimed the chip beat an nvidia rtx 5080 by more than 3x on deepseek r1 inference. a lunchbox sized pc outrunning a $1,000 discrete gpu on a real ai workload

a heavy ai user pays $200 for claude code max, $200 for chatgpt pro, $20 for cursor and $20 for gemini. that's $5,280 a year and the box pays itself off in 9 to 10 months

install ollama, pull the model, point claude code at localhost. same interface, nothing leaves the machine, nothing costs per request

bookmark this and read the article below

[Thad Scheer @ThadOfSphere]

When the ordinary person has the tools to build and field weapons of mass destruction from their bedroom, you get national security policies like this. Nonsensical yet oddly necessary.

[Anthropic @AnthropicAI]

The US government, citing national security authorities, has issued an export control directive to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign national Anthropic employees.

The net effect of this order is that we must abruptly disable Fable 5 and Mythos 5 for all our customers to ensure compliance.

Access to all other Claude models is not affected.

We apologize for this disruption to our customers. We believe this is a misunderstanding and are working to restore access as soon as possible.

Read our full statement: https://t.co/bwn0sximKZ

[Thad Scheer @ThadOfSphere]

Jeff Bezos is digging in the right place.

Backpropagation is how artificial neural networks learn. It's monstrously inefficient in both energy and hardware. Animals do not use backpropagation and our brains run on watts, not gigawatts.
https://t.co/d3l6bwZq3H

[Thad Scheer @ThadOfSphere]

Headline: FCC wants to Kill Burner Phones

Reality: This is about countering nation-state SIM farms that can attack our cell networks with denial of service. Think NYC near the UN during the UNGA. Inactive farms are hard to find/stop.

https://t.co/f1D2hxl8oT

[Thad Scheer @ThadOfSphere]

@StuartHameroff Current physics repeats this line, that the external observer is not needed to collapse the wavefunction. This is accepted. But is it true?

Above my pay grade.

[Ben Dickson @bendee983]

Anthropic designing Mythos/Fable to prevent frontier AI research is the most disappointing part of their release. Hope it doesn't set a precedent (I don't think it will).

But what it will surely do is push serious researchers/organizations to use open weight LLMs instead of using Anthropic products.

Absolutely unnecessary move. Own goal.

[Thad Scheer @ThadOfSphere]

Great post. The enemy is an idea, not a system. But the idea has always had a delivery system. The Soviets realized the West could only be defeated through internal organ failure, not blunt force. They promoted the French idea to "useful idiots", i.e., academics and activists. The Soviets were the invisible organizing force behind the anti-nuclear movement of the 70s-80s, then later the climate movement. The Soviets were behind all the activist movements of the late 20th century. Beijing took over where Moscow left off, cultivating more useful idiots, i.e., multi-national corporations (particularly entertainment and news), politicians (bought and paid for or morally compromised), and social media. The "idea" is no longer confined to academics and activists, it has fully metastasized in Western culture. As a free-ish society, we have no natural antibodies, no defense. The idea spreads to every organ.

But there is a system delivering it. What started as a psyop in Moscow is now run by Beijing, with help from European elitists. The system is actively promoting multiple organ failure (MOF) in the West, disabling antibodies before they can organize.

People on the left are naturally susceptible to the [French] idea and receive it easily, but they aren't the only useful idiots. People on the right are equally as bad, easily bated into activism opposing their fellow citizens on the left, as if liberals are the enemy. People on the right are gullible in the presence of ideology and easily driven to pointless activism. They argue with family, friends, and neighbors when they should be focusing on the delivery system, not the victims.

One cannot fight a nameless shapeless idea, but one can fight a system. Foreign influence is the delivery system that must be shut down. Foreign influence must be investigated and excised. This is difficult because both media and politicians are compromised. Investigations with teeth never get going, and private investigations get discredited. Beijing's reach is embedded in the institutions. Citizen journalists and free speech media are the only option.

If the average American "activist" who is driven to rage by issue-politics (on the right or left) were to learn the degree to which their passion and outrage was the carefully designed product of a foreign intelligence service's targeted psyop, they might look closer at their passions and issues. Most of these activating issues don't hold up under scrutiny. Simply awakening our own people is step #1.

Kill the delivery system of the [French] idea, then we can neutralize the idea itself. It's not a very good idea, so we have chance.

[Thad Scheer @ThadOfSphere]

@RodDMartin How certain are you that China has ~1.4b population today, or anywhere close to that figure?

[Thad Scheer @ThadOfSphere]

@dhinchcliffe This is the correct (and best) take on this. Your advice is 100% on.

[Thad Scheer @ThadOfSphere]

@StuartHameroff It might take > 1 beer, but I wish @anilkseth would do it. I sense he's taking this convo as more adversarial than it really is. These are the biggest unanswered questions in science. Worthy of 2 beers.

As Anil has said several times, you guys are basically on the same side.