That’s fair. I see what I see at an engineering and architecture level. You see what you see at the business level.

That said. I stand by my statement because I and most of my colleagues in similar roles get continued, repeated and expanded-scope engagements. Definitely in LLMs and genAI in general especially over the last 3-5 years or so, but definitely not just in LLMs.

“AI” is an incredibly wide and deep field; much more so than the common perception of what it is and does.

Perhaps I’m just not as jaded in my tech career.

operations research, and conventional software which never makes mistakes if it’s programmed correctly.

Now this is where I push back. I spent the first decade of my tech career doing ops research/industrial engineering (in parallel with process engineering). You’d shit a brick if you knew how much “fudge-factoring” and “completely disconnected from reality—aka we have no fucking clue” assumptions go into the “conventional” models that inform supply-chain analytics, business process engineering, etc. To state that they “never make mistakes” is laughable.

Absolutely not true. Disclaimer, I do work for NVIDIA as a forward deployed AI Engineer/Solutions Architect—meaning I don’t build AI software internally for NVIDIA but I embed with their customers’ engineering teams to help them build their AI software and deploy and run their models on NVIDIA hardware and software. edit: any opinions stated are solely my own, N has a PR office to state any official company opinions.

To state this as simply as possible: I wouldn’t have a job if our customers weren’t seeing tremendous benefit from AI technology. The companies I work with typically are very sensitive to CapX and OpX costs of AI—they self-serve in private clouds. If it doesn’t help them make money (revenue growth) or save money (efficiency), then it’s gone—and so am I. I’ve seen it happen; entire engineering teams laid off because a technology just couldn’t be implemented in a cost-effective way.

LLMs are a small subset of AI and Accelerated-Compute workflows in general.

That was no steakhouse. You straight up walked into a Mafia front. Like out of Michael Jackson’s Smooth Criminal music video.

All you need is a brilliant high school student with a knack for nuclear physics, a rebellious streak, and a desire to prove their intellectual superiority. As shown in the 1986 documentary, The Manhattan Project, even a resourceful teen can assemble a nuclear weapon in a matter of weeks if they manage to access the right materials. Just look for someone with the right mix of ambition, genius, and a little disregard for the rules, and you’ll have yourself a homemade nuclear device in no time!

I like to splurge on a Shackshackshack from time to time.

We’re looking at this from opposite sides of the same coin.

The NN graph is written at a high-level in Python using frameworks (PyTorch, Tensorflow—man I really don’t miss TF after jumping to Torch :) ).

But the calculations don’t execute on the Python kernel—sure you could write it to do so but it would be sloooow. The actual network of calculations happen within the framework internals; C++. Then depending on the hardware you want to run it on, you go down to BLAS or CUDA, etc. all of which are written in low-level languages like Fortran or C.

Numpy fits into places all throughout this stack and its performant pieces are mostly implemented in C.

Any way you slice it: the post I was responding to is to argue that AI IS CODE. No two ways about that. It’s also the weights and biases and activations of the models that have been trained.

deleted by creator

Neural nets are typically written in C; then frameworks abstract on top of that (like Torch, or Tensorflow) providing higher-level APIs to languages like (most commonly) Python, or JavaScript.

There are some other nn implementations in Rust, C++, etc.

I don’t care. Will always be Nome. Fuck Pedo Stallman’s preference.

Just like our favorite DC supervillain, Bane.

Wet Cougar in the bathtub.

Costco’s soft-serve is way better than McD’s and actually is cheap.

Bullshit. Developers never make mistakes. N.E.V.R.

Removed by mod

No, but it’s basically a “I can use it to build my billion-dollar business and keep the profits if I want” license. The only real catch is that if I decide to modify the code and distribute it, I’m required by the license to share those changes with whoever gets the modified version. There’s nothing in the GPL that stops me from being a downstream freeloader, and I can stay on whatever version I like—no one’s forcing me to update to newer ones with terms I don’t agree with. Forking and modifying for my own needs is totally fine, as long as I slap the same GPL on the changes if I hand them out.

1:0

You can scan before the encryption step. It defeats the purpose of the encryption such that only the privileged actor gets plaintext while everyone downstream gets encrypted bytes, but technically it’s possible.

It’s only a matter of time until a vulnerability in the privilege is found and silently exploited by a nefarious monkey, and that’s precisely why adding backdoors should never be done.

I’d say it actually goes further. We have plenty of evidence leading to the realization of fact that simply measuring a phenomenon changes the phenomenon. From a quantum mechanics perspective we say things like “measuring the phenomenon collapses its wave function to a single state.”

When a quantum system is measured, its wave function, which represents a superposition of multiple potential outcomes, collapses to a single definite state corresponding to the result of the measurement.

All macroscopic phenomena comprise nanoscopic quantum phenomena.

Super fucking weird to think about. The classic undergrad physics experiment is the double-slit experiment— particles like electrons create an interference pattern when unobserved, acting like waves and passing through both slits at once. However, when we measure which slit a particle goes through, this wave-like behavior disappears, and the particle behaves as if it went through only one slit. This shows that measurement collapses the particle’s wave function from multiple possibilities into a single, definite state.

Similarly, despite being depicted as such in early exposures to chemistry, electrons don’t “orbit” the nucleus like planets do their stars—rather they have regions around the nucleus in which they are more probably found. These misleadingly named “orbitals” vary in shape.

Finally, we have the Heisenberg Uncertainty Principle; which states that we can measure either a particle’s speed (kinetic energy) or its location, but not both, because the act of measuring (observing) that particle irrevocably changes it.

Here’s a macroscopic example of how measuring/observing things changes the thing. When you measure the temperature of an object using a thermometer, the object is either transmitting or receiving thermal energy to/from the thermometer, because the thermometer needs to be in contact and thermal equilibrium with the object. The object’s total energy level has now changed—even if it’s a trivial change it’s also non-zero. Measuring/observing the object in this way has changed it.

omg it goes deeper. I love physics. Classical mechanics models work well when we want to explain and predict macroscopic and limited chain-of-events phenomena. We can predict with high confidence that a 2000 kg car traveling at 100 km/h will impulse this much force and energy to a stationary object when they collide, assuming a perfectly inelastic collision, spherical cows, etc. We can’t model with any confidence with any classical model how the displaced air molecules from this collision in Nuremberg, Germany will create tornadoes in six months in Wichita, Kansas, USA. That’s the butterfly effect.

Ultimately, this interplay between measurement and outcome highlights a fundamental truth in both quantum mechanics and chaos theory: the universe is inherently unpredictable at every scale. Just as the behavior of subatomic particles is influenced by the act of observation, the butterfly effect shows us that small changes can lead to significant consequences in complex systems. This intertwining of uncertainty and complexity underscores the limitations of our predictive models, whether they pertain to the quantum realm or the macroscopic world.

I’m not a political scientist or at all a political expert, so I lack the background to make a supported or definitive statement to that effect. So this is my opinion based on what I’ve seen and read but is otherwise unprofessionally qualified.

Most people probably already know how they would vote if polling day was today. The issue as I see it is not so much about chasing after the undecided voters as much as it driving engagement and enthusiasm upwards such that people actually will go vote. It’s hard to drive that enthusiasm and engagement without facetime, without being on stage giving speeches about the issues, without debating about policies and implementation. With the win/loss margins being so razor thin right now in some of the key battleground states (so-called “swing states”) anything that gets her away from the voters is a distraction that can’t be allowed to manifest.

If she wins the election being the key operative word. The race is far from won right now; her attention needs to be on her race. Ukraine does not benefit in the long run if she loses it now.