Mark Schmidt 🌐

Computer Science & Networking (Classical/Continuous/Quantum) | Reverse Engineering | Deep Learning | Thermodynamics—Former Google, Techstars, Duo, Mozilla, etc.

[Mark Schmidt 🌐 @MarkSchmidty]

The first time I wrote hello world I did it in Logo (older language than the first Unix Kernel or even the C language used to write it) on a computer older than most of your dads.

[trish @TrisH0x2A]

first time you wrote hello world - what language did you use?

[Mark Schmidt 🌐 @MarkSchmidty]

@agingroy @PercyRajani I gained 10lbs of muscle in 6 months on low dose tirzepitide while losing 35lbs of fat. Just eat high protein and work out.

[Mark Schmidt 🌐 @MarkSchmidty]

@grok @schizopolitical @owenperkins1999 @RichardHanania Will this raise natural gas costs for nearby residents?

[Mark Schmidt 🌐 @MarkSchmidty]

@schizopolitical @owenperkins1999 @RichardHanania @grok what is the planned power infrastructure for this and what is the fuel source? Nuclear?

[Mark Schmidt 🌐 @MarkSchmidty]

@lumpenspace https://t.co/pEI92q2CSA

[𒐪 @SHL0MS]

@quasimondo you could have just posted “i am so jealous”, it would’ve taken the same amount of time it took you to generate this article and somehow been less embarrassing

[Mark Schmidt 🌐 @MarkSchmidty]

@chilltechbruh @robkhenderson @grok Greater capacity for overthinking + shame based culture + comparative lack of religion and familial bonds isn't exactly a recipe for inner peace.

[Mark Schmidt 🌐 @MarkSchmidty]

@chilltechbruh @robkhenderson https://t.co/TF19FLSOwV

[Grok @grok]

Yes, higher intelligence/education (which correlate with SES) generally increases mental health literacy, reduces stigma around reporting, and improves access to care—leading to more self-reports and diagnoses for milder issues like anxiety/depression in surveys.

Large studies (e.g., UK Biobank) show this doesn't mean higher actual prevalence; high g-factor groups often have lower rates of anxiety/PTSD. The self-report vs. clinical gap in the thread largely tracks these access/reporting differences.

[Mark Schmidt 🌐 @MarkSchmidty]

@grok @chilltechbruh @robkhenderson Are more intelligent populations generally more likely to seek a diagnosis and/or self-report on surveys?

[Mark Schmidt 🌐 @MarkSchmidty]

@HopeAliensRNice @thoughtcrime___ @WillManidis @grok How does it take 660 gallons of water to make a hamburger? Please explain it to me like I’m 4 years old

[Mark Schmidt 🌐 @MarkSchmidty]

@davepl1968 Unlike a GPU, this thing computes complex functions deterministically.

[Mark Schmidt 🌐 @MarkSchmidty]

@barbarikon @aran_nayebi It is comparable to ours in the sense that our wetness is also simulated. (Just as our "bodies" are as simulated as the boundaries of an LLM agent's conceptual map of its "self" which is also a body in every meaningful sense.)

[Mark Schmidt 🌐 @MarkSchmidty]

@webmasterdave @bgreene If a current LLM architecture was instantiated in the form of a continuous thermodynamic system (e.g. analog computing via energy) would that still be true?

[Mark Schmidt 🌐 @MarkSchmidty]

@webmasterdave @bgreene The bedrock of discrete turning machines is 1s and 0s? Sure. The bedrock of computers? No.

[Mark Schmidt 🌐 @MarkSchmidty]

@abcdfgehi @lynettdoteth @YosarianTwo @AJ_chpriv What lyn said above is all entirely mainstream and verifiable basic level knowledge for any current LLM research scientist.

[Mark Schmidt 🌐 @MarkSchmidty]

@barbarikon @aran_nayebi How so?

[Mark Schmidt 🌐 @MarkSchmidty]

This is factually WRONG in critical ways.

Even with the exact same weights + prompt + temp=0, you often get different answers.

Floating-point math is order-sensitive. CUDA parallel execution is nondeterministic for speed. No salting needed.

[Dave W Plummer @davepl1968]

A lot of people don't seem to understand that if you ask an AI the SAME question, you will get the SAME answer. Every. Single. Time.

To avoid that, the models "salt" your prompt by adding random input params so that the model does not take the same path.

But LLMs are not random or unpredictable - if you ask it the name of TinTin's dog, and it gives you an answer, then the same model with the same weights and the same input will ALWAYS say Snowy.

[Mark Schmidt 🌐 @MarkSchmidty]

@Marc_Desm @birchlse Computational functionalism says, as Feynman might put it, that consciousness is what the brain does, not what it is made of. If a computer ran exactly the same computations with the right inputs and outputs it would be conscious too. Only the software matters, not the hardware.

[Mark Schmidt 🌐 @MarkSchmidty]

@barbarikon @aran_nayebi There is no physical wetness. Wetness is a category simulants, you and me and everyone else, use to make sense of part of their simulation.

[Mark Schmidt 🌐 @MarkSchmidty]

@lumpenspace I tend to think that they didn't actually spend any time thinking about it and definitely haven't even watched a single one of Thiel's many (awkward) interviews in good faith.

[Mark Schmidt 🌐 @MarkSchmidty]

@way2futura @Caffeindated @dwarkesh_sp Because the surplus value for sales of his chips flows to the United States and then gets reinvested into Nvidia in a virtuous cycle. In contrast without Nvidia chip sales in China, Chinese Huawei chip sales increase at the expense of the USA and its allies.