Chances of being caught are close to zero (I have contacted many times authors of papers who's work I was unable to replicate - most of the time zero reply, sometimes "yeah it was a honest mistake, oops"), super high competition (only a few tenured positions in all world's high visibility institutions per year), full control over student's future and being able to force them to do fraud (and later blame on them).

Obviously, not all, blah blah - but many academic scientists are the last people that should be doing science.

I grew up in Eastern Europe (Warsaw) in "commie" blocks and there was a lot of valid criticisms and problems (like poor quality of buildings, small apartments, or thin walls - but consider that they rebuilt whole Warsaw after it was completely grounded in WW2 in a decade or two!), but also a lot to love. Extremely walkable, safe, all amenities (cinemas, stores, cultural centers, playgrounds) in the walking distance, lots of trees and green, easy access to public transit. As a kid or teenager they were great. I preferred it 100x over suburbs where my parents moved later, and to typical American cityscapes. (And this is why I moved to NYC and love it)

Here is a fun and a bit provocative/exaggerated video https://youtu.be/1eIxUuuJX7Y

Everyone is different so I'm not forcing my perspective onto anyone, just worth considering - especially if you have not had such first hand experience (and the main objection to tenements comes from how depressing they look or American association of "projects = crime", which misses a lot of "why"). Feel free to disagree!

And apart from that, I don't think "more of small houses" solves anything. It has to create more car dependence and social isolation. And it does not really scale, where would you fit more of smaller homes in SF?

I think this is the root of it. You prefer higher density -- and that's great. I'm sure not everyone agrees, but I don't see any reason to take that away. In fact, I think it should be encouraged for those that like it.

The issue, IMHO, is that some folks don't like that and prefer lower density. And a lot of these changes focus on taking that away from them (i.e. changing their current neighborhood).

Also, just a comment on: "It has to create more car dependence and social isolation"

I don't think that's true. I live in a pretty traditional SFH neighborhood. Within a 12 minute bike ride, I have:

* Four grocery stores (Major chains)

* 2 gyms

* Dozens of restaurants

* Several large parks

* 2 home improvement stores

* Several large employers

* Several (non-Starbucks) coffee shops

And lots more. It's certainly possible, with bikes, to have SFH neighborhoods where cars aren't required.

Many of the people who are pissed about more density coming to cities only moved to them in the last decade or two, especially on the east coast where white flight only recently reversed.

But my question remains - how do you scale up your approach to the already-full SF? How do you make it more affordable, as prices are insane due to demand >> supply? Or do you just envision a more sprawled, but similarly dense SF as the solution?

I agree with a lot, but like others - disagree with some.

My background: I've worked in tech for ~14 in all kinds of roles - from pure junior IC, through a "team lead" (something between an expert IC, a tech lead, and a manager), "tech lead," company's "technical architect" (highest level tech lead, peer to the technical director, but without any direct reports), and something akin to a tech director. Now I'm back to IC. Companies from small gamedev ones (80 people total, 15 engineers), medium gamedev (30 engineers and coding technical artists), huge gamedev (Ubisoft where you can have 100+ engineers and 1500 people total on a project), and for the last 7y "big tech".

The idea I would like to push back the most is that "your words have more weight". I have never had trouble getting my opinions heard, even if I didn't push them. Being the expert IC and "problem solver" sometimes I'd have even CEO asking me directly for advice and how to solve some issues (both technical and non-technical!). Not always following them, but I didn't expect that. Having an official title, in theory, you can use some "authority" to formally push those ideas. But... in practice, it does not work better. People will still go directly to the most technical experts. And if you abuse the position/authority/title (I hope I never did that, but that's not for me to judge...), it can cause resentment, pushback, and more disagreements. You will also hear less of gossip and honest feedback, to some engineers you becomes "not one of us anymore".

It can also destroy friendships. I had a great friend (meeting socially with our wives once a week, sharing interests) who was my peer and was then promoted to my lead. I still really liked them and wanted to stay friends. We always had technical disagreements (which were fine for peer ICs). Later, some of those technical disagreements and my bringing up issues publicly caused him to get bitter with me (as the upper management saw those and took my side on some occasions), and eventually stopped the friendship completely; after I left the company, they started ghosting me. :(

Similarly, on a few occasions, I agreed to lead/manage formally (in one case, it came from me - in other cases, I was asked to). I agreed because I thought, "Things are f-d up; I can solve them by being closer to the upper leadership and helping the team succeed." Man, I was so naive. :( I didn't have any more authority or power with the higher-ups, and there were more disagreements. They expected me to enforce policies I disagreed with. As you can imagine, this didn't last long, and I always ended up leaving the team/company and being super burnt out.

So now I'm happy to be a staff-level IC, an expert, and a "hacker," playing with problems hands-on and building my expertise further. The field grows so quickly that there is always something exciting and new to learn and do. I would happily be a tech lead of some project close to me (luckily, at Google and similar, it's flexible, per-project, and not formal), but I probably do not want to manage again. Maybe it will change, depends.

It's used for quickly hacked debug tools to interleave UI and regular logic and not do a logic/view separation (as it would result in code bloat and a necessity for a refactor). You want UI code do some logic (like modifying properties of some game entity or renderer) and prefer to inline it. Lots of game code is effectively YOLO without even a single test. It's also typically guarded by IfDefs and compiled out of release versions.

But as soon as it stops being just hacky debuggers and people try to write proper tools in it, it becomes much more of a pain - people try to (poorly) emulate a retained mode in it, hold state, cache - and it becomes unreadable mess.

Effectively people are hasty and don't spend the time to try doing things nicely, in particular because the first steps and debug use allow you to do quick things.

But I don't think it's a fundamental property of IMGUI or Dear ImGui that "proper tools" become particularly more of a pain. Of course it is more work to make a proper tools than hasty-debug-tools, and I can easily see how underengineering can back-fire (over-engineering it likewise).

This is how almost every research publication in every field - from medicine, through psychology, to computer science looks like.

Maby startups don't even need to take on seed investor money (could be self funded, smaller team, longer time before funding etc) - but founders obviously prefer to have a nice salary and soft landing. Or sometimes literally just as a marketing strategy "we are backed by x/y/z, this proves our value". This is the price to pay. What am I missing?

Until then, it's a bit funny claim, especially considering what a failure OpenCL was (programmer's experience and fading support). Or trying to do GPGPU with compute shaders in DX/GL/Vulkan. Are they really "motivated"? Because they had so many years and the results are miserable... And I don't think they invested even a fraction of what got invested into CUDA. Put your money where your mouth is.

But what you mention is important, and also a reason for the ultimate demise of e.g. Xeon Phi. Intel surely realized they could just scale their existing Xeon designs up-and-out further than expected. Like from a product/SKU standpoint, what is the point of having a 300 core Phi where every core is slow as shit, when you have a 100 core 4-socket Xeon design on the horizon, using an existing battle-tested design that you ship billions of dollars worth every year? Especially when the 300 core Xeon fails completely against the competition. By the time Phi died, they were already doing 100-cores-per-socket systems. They essentially realized any market they could have had would be served better by the existing Xeon line and by playing to their existing strengths.

This made me wonder a couple of things-

What kind of workloads and problems is that best suited for? It’s a lot of cores for a CPU, but for pure math/compute, like with AI training and inference and with graphics, 288 cores is like ~1.5% of the number of threads of a modern GPU, right? Doesn’t it take particular kinds of problems to make a 288 core CPU attractive?

I also wondered if the ratio of the highest core count CPU to GPU has been relatively flat for a while? Which way is it trending- which of CPUs or GPUs are getting more cores faster?

Here is a relevant article: https://www.kdnuggets.com/2020/03/deep-learning-breakthrough...

And what you suggest could be done, but it would likely flop commercially if you made it today, which is why they aren’t doing it. SIMD machines are faster on homogenous workloads, by a lot. It would be a bummer to develop a CPU with thousands of cores that is still tens or hundreds of times slower than a comparably priced GPU.

SIMD isn’t going away anytime soon, or maybe ever. When the workload is embarrassingly parallel, it’s cheaper and more efficient to use SIMD over general purpose cores. Specialized chiplets and co-processors are on the rise too, co-inciding with the wane of Moore’s law; specialization is often the lowest hanging fruit for improving efficiency now.

There’s going to be plenty of demand for general programmers but maybe worth keeping in mind the kinds of opportunities that are opening up for people who can learn and develop special purpose hardware and software.

If you don't want that, program the GPU directly in assembly or C++ or whatever. A kernel is a thread - program counter, register file, independent execution from the other threads.

There isn't a Linux kernel equivalent sitting between you and the hardware so it's very like bare metal x64 programming, but you could put a kernel abstraction on it if you wanted.

Core isn't very well defined, but if we go with "number of independent program counters live at the same time" it's a few thousand.

X64 cores are vaguely equivalent to GCN compute units, 100 or so if either in a 300W envelope. X64 has two threads and a load of branch prediction / speculation hardware. GCN has 80 threads and swaps between them each cycle. Same sort of idea, different allocation of silicon.

Traditional compute? The cores were too weak.

Number crunching? Okay-ish but gpus were better.

Useless stuff.

I'd love to hear about what didn't work. OpenMP support seemed ok maybe but OpenMP is just a platform, figuring out software architectures that's mechanistically sympathetic to the system is hard. It would be so interesting to see what Xeon Phi might have been if we had Calcite or Velox or OpenXLA or other execution engine/optimizers that can orchestrate usage. The possibility of something like Phi seems so much higher now.

There's such a consensus around Phi tanking, and yes, some people came and tried and failed. But most of those lessons, of why it wasn't working (or was!) never survived the era, never were turned into stories & research that illuminates what Phi really was. My feeling is that most people were staying the course on GPU stuff, and that there weren't that many people trying Phi. I'd like more than the heresay heaped at Phi's feed to judge by.

Math guys came up with a list of algorithms to try for a search engine backend.

What we needed was matrix multiplication and maybe some decision tree walking (that was some time ago, trees were still big back then, NNs were seen as too compute-intensive for no clear benefits). So we thought that it might be cool to have a tool that would support both. Phi sounded just right for both.

And things written to AVX-512 did work. Software surpisingly easy to port.

But then comes the usual SIMD/CPU trouble: every SIMD generation wants a little software rewrite. So for both Phi generations we had to update our code. For things not compatible with the SIMD approach (think tree-walking) it is just a weak x86.

In theory Phi's were universal, in practice what we got was: okay number crunching, bad generic compute.

GPU was somewhat similar: the software stack was unstable, CUDA just did not materialize as a standard yet. But every generation introduced a massive increase in compute available. And boy did NVIDIA move fast...

So GPU situation was: amazing number crunching, no generic compute.

And then there were a few ML breakthroughs results which rendered everything that did not look like a matrix multiplication obsolete.

PS I wouldn't take this story too seriously, details may vary.

- Very bad performance at existing x86 workloads, so a major selling point was basically not there in practice, because extracting any meaningful performance required a software rewrite anyway. This was an important adoption criteria; if they outright said "All your existing workloads are compatible, but will perform like complete dogshit", why would anyone bother? Compatibility was a big selling point that ended up meaning little in practice, unfortunately.

- Not actually what x86 users wanted. This was at the height of "Intel stagnation" and while I think they were experimenting with lots of stuff, well, in this case, they were serving a market that didn't really want what they had (or at least wasn't convinced they wanted it).

- GPU creators weren't sitting idle and twiddling their thumbs. Nvidia was continuously improving performance and programmability of their GPUs across all segments (gaming, HPC, datacenters, scientific workloads) while this was all happening. They improved their compilers, programming models, and microarchitecture. They did not sit by on any of these fronts.

Ironically the main living legacy of Phi is AVX-512, which people did and still do want. But that kind of gives it all away, doesn't it? People didn't want a new massively multicore microarchitecture. They wanted new vector instructions that were flexible and easier to program than what they had -- and AVX-512 is really much better. They wanted the things they were already doing to get better, not things that were like, effectively a different market.

Anyway, the most important point is probably the last one, honestly. Like we could talk a lot about compiler optimizations or autovectorization. But really, the market that Phi was trying to occupy just wasn't actually that big, and in the end, GPUs got better at things they were bad at, quicker than Phi got better at things it was bad at. It's not dissimilar to Optane. Technically interesting, and I mourn its death, but the competition simply improved faster than the adoption rate of the new thing, and so flash is what we have.

Once you factor in that you have to rewrite software to get meaningful performance uplift, the rest sort of falls into place. Keep in mind that if you have a $10,000 chip and you can only extract 50% of the performance, you more or less have just $5,000 on fire for nothing in return. You might as well go all the way and use a GPU because at least then you're getting more ops/mm^2 of silicon.

I tend to think there was tons of untapped potential still on the table. And that a failure to adopt potential isn't purely Intel alone's fault. The story we are commenting on is about the rest-of-industry trying to figure out enduring joint strategies, and much of this is chipmaker provided, but it is also informed and helped by plenty of consumers also pouring energy in to figure out what's working and not, trying to push the bounds.

Agreed that anyone going in thinking Xeon Phi would be viable for running a boring everyday x86 workload was going to be sad. To me the promise seemed clear that existing toolchains & code would work, but it was always clear to me there were a bunch of little punycores & massive SIMD units and that doing anything not SIMD intensive wasn't going to go well at all. But what's the current trend? Intel and AMD are both actively building not punycores but smaller cores, with Sierra Forest and Bergamo. E-cores are the grown up Atom we saw here.

Yes the GPGPU folks were winning. They had a huge head start, were the default option. And Intel was having trouble delivering nodes. So yes, Xeon Phi was getting trounced for real reasons. But they weren't architectural issues! It just means the Xeon Phi premise was becoming increasingly handicapped.

As I said I broadly agree everywhere. Your core point about giving the market more of what it already does is well taken, is a river of wisdom we see again and again. But I do think conservative thinking, iterating along, is dangerous thinking that obstructs us from seeing real value & possibility before us. Maybe Intel could have made a better ML chip than the GPGPU market has gotten for years, had things gone differently; I think the industry could perhaps have been glad they had veered onto a new course, but the barriers to that happening & the slow down in Intel delivery & the difficulty bootstrapping new software were all horrible encumberances which were rightly more than was worth bearing together.

Most experimenters saw it as a way to have something GPU-like in terms of raw power but with no limitations charateristic of SIMT's. Like, slightly different code paths for threads doing number crunching or something.

But it turns out that it's easier to force everything into a matrix. Or a very big matrix. Or a very-very-very big matrix.

And then see what sticks.

When CPUs actually come with bandwidth and a decent vector unit, such as the A64FX, lo and behold, they lead the Top500 supercomputer list, also beating out GPUs of the day.

Why have we not been getting bandwidth in CPUs? Is it because SPECint benchmarks do not use much? Or because there is too much branch-heavy code, so we think hundreds of cores are helpful?

Existing machines are ridiculously imbalanced, hundreds of times more compute vs bandwidth than the 1:1 still seen in the 90s. Hence matmul as a way of using/wasting the extra compute.

The AMD MI300a looks like a very interesting development: >5 TB/s shared by 24 cores plus GPUs.

In practice, people weren’t afraid to roll up their sleeves and write CUDA code. If you wanted good performance you had to think about data parallelism anyways, and at that point you’re not benefiting from x86 backwards compatibility. It was a fascinating dream while it lasted though.

There's no intrinsic reason why multiplying matrices requires massive parallelism, in principle it could be done on few cores plus good management of ALUs/memory bandwidth/caches.

[1]: https://github.com/bevyengine/bevy/blob/main/examples/shader...

> setting up the graphics pipe

I’ve picked D3D11, as opposed to D3D12 or Vulkan. The 11 is significantly higher level, and much easier to use.

> compiling, binding

The compiler is design-time, I ship them compiled, and integrated into the IDE. I solved the bindings with a simple code generation tool, which parses HLSL and generates C#.

> No proper debugging

I partially agree but still, we have renderdoc.

Indeed, in D3D they are called “wave intrinsics” and require D3D12. But that’s IMO a reasonable price to pay for hardware compatibility.

> no cooperative matrix multiplication

Matrix multiplication compute shader which uses group shared memory for cooperative loads: https://github.com/Const-me/Cgml/blob/master/Mistral/Mistral...

> tensor cores

When running inference on end-user computers, for many practical applications users don’t care about throughput. They only have a single audio stream / chat / picture being generated, their batch size is a small number often just 1, and they mostly care about latency, not throughput. Under these conditions inference is guaranteed to bottleneck on memory bandwidth, as opposed to compute. For use cases like that, tensor cores are useless.

> there's no way D3D11 can compete with either CUDA

My D3D11 port of Whisper outperformed original CUDA-based implementation running on the same GPU: https://github.com/Const-me/Whisper/

Most ML inferencing is throttled with memory, not compute. This certainly applies to both Whisper and Mistral models.

> it certainly doesn't spell the end of CUDA

No, because traditional HPC. Some people in the industry spent many man-years developing very complicated compute kernels, which are very expensive to port.

AI is another story. Not too hard to port from CUDA to compute shaders, because the GPU-running code is rather simple.

Moreover, it can help with performance just by removing abstraction layers. I think the reason why compute shaders-based Whisper outperformed CUDA-based version on the same GPU, these implementations do slightly different things. Unlike Python and Torch, compute shaders actually program GPUs as opposed to calling libraries with tons of abstractions layers inside them. This saves memory bandwidth storing and then loading temporary tensors.

One of the features that would get compute shaders far ahead compared to now would be pointers and pointer casting - Just let me have a byte buffer and easily cast the bytes to whatever I want. Another would be function pointers. These two are pretty much the main reason I had to stop doing a project in OpenGL/Vulkan, and start using CUDA. There are so many more, however, that make life easier like cooperative groups with device-wide sync, being able to allocate a single buffer with all the GPU memory, recursion, etc.

Khronos should start supporting C++20 for shaders (basically what CUDA is) and stop the glsl or spirv nonsense.

So yet another thing where Khronos APIs are dependent on DirectX evolution.

It used to be that AMD and NVidia would first implement new stuff on DirectX in collaboration with Microsoft, have them as extensions in OpenGL, and eventually as standard features.

Now even the shading language is part of it.

Modern Vulkan is looking pretty good now. Cooperative matrix multiplication has also landed (as a widely supported extension), and I think it's fair to say it's gone past OpenCL.

Whether we get significant adoption of all this I think is too early to say, but I think it's a plausible foundation for real stuff. It's no longer just a toy.

[1] https://community.arm.com/arm-community-blogs/b/graphics-gam...

It's been awesome seeing folks like Keras 3.0 kicking out broad Intercompatibility across JAX, TF, Pytorch, powered by flexible executuon engines. Looking forward to seeing more Vulkan based runs getting socialized benchmarked & compared. https://news.ycombinator.com/item?id=38446353

[1]: https://kompute.cc/

[2]: https://github.com/webonnx/wonnx

In researching this answer, I came across a really interesting thread[1] on diagnosing performance problems with USM in SYCL (running on AMD HIP in this case). It's a good tour of why this is hard, and why for the vast majority of users it's far better to just use CUDA and not have to deal with any of this bullshit - things pretty much just work.

When targeting compute shaders, you pretty much have to manage buffers manually, and also do copying between host and device memory explicitly (when needed - on hardware such as Apple Silicon, you prefer to not copy). I personally don't have a problem with this, as I like things being explicit, but it is definitely one of the ergonomic advantages of modern CUDA, and one of the reasons why fully automated conversion to other runtimes is not going to work well.

[1]: https://stackoverflow.com/questions/76700305/4000-performanc...

SYCL builds on top of OpenCL so you need to know the history of OpenCL. OpenCL 2.0 introduced shared virtual memory, which is basically the most insane way of doing it. Even with coarse grained shared virtual memory, memory pages can transparently migrate from host to device on access. This is difficult to implement in hardware. The only good implementations were on iGPUs simply because the memory is already shared. No vendor, not even AMD could implement this demanding feature. You would need full cache coherence from the processor to the GPU, something that is only possible with something like CXL and that one isn't ready even to this day.

So OpenCL 2.x was basically dead. It has unimplementable mandatory features so nobody wrote software for OpenCL 2.x.

Khronos then decided to make OpenCL 3.0, which gets rid of all these difficult to implement features so vendors can finally move on.

So, Intel is building their Arc GPUs and they decided to create a variant of shared virtual memory that is actually implementable called unified shared memory.

The idea is the following: All USM buffers are accessible by CPU and GPU, but the location is defined by the developer. Host memory stays on the host and the GPU must access it over PCIe. Device memory stays on the GPU and the host must access it over PCIe. These types of memory already cover the vast majority of use cases and can be implemented by anyone. Then finally, there is "shared" memory, which can migrate between CPU and GPU in a coarse grained matter. This isn't page level. The entire buffer gets moved as far as I am aware. This allows you to do CPU work then GPU work and then CPU work. What doesn't exist is a fully cache coherent form of shared memory.

https://registry.khronos.org/OpenCL/extensions/intel/cl_inte...

> [1] https://community.arm.com/arm-community-blogs/b/graphics-gam...

"Using a pointer in a shader - In Vulkan GLSL, there is the GL_EXT_buffer_reference extension "

That extension is utter garbage. I tried it. It was the last thing I tried before giving up on GLSL/Vulkan and switching to CUDA. It was the nail in the coffin that made me go "okay, if that's the best Vulkan can do, then I need to switch to CUDA". It's incredibly cumbersome, confusing and verbose.

What's needed are regular, simple, C-like pointers.

Predictably?, amusingly? police never had access to this data, until a government minister was murdered in 2003, when a sample from the suspect was retrieved. From what we know it has not been used since. So we can be cynical, but under the circumstances, the police use of the registry has not yet taken hold and is guarded by the courts..

But assuming they obey the law they did not used your samples.

So there are privacy laws in place? Also they could have been cleaning old results/samples and this was one step.

And also - could eg. police use it?

> They already did something with my biological samples

I can only guess you were tested for something in the hospital. Samples are sent to the lab (separate department) to be tested. If additional tests need to be perform they can use the blood they already received. The samples are kept for ready availability if additional tests are requested by doctor. Doctors dont care how its done, they dont have time to inform lab patient x is out home.

After some time they need to be destroyed - due to expiry date on it. Before destroying the lab contacted you and asked for dna permission.

> And also - could eg. police use it?

I don't know that. But you might want to check how medical data is protected in your jurisdiction.

See https://www.hhs.gov/ohrp/regulations-and-policy/guidance/faq... under "Should the initial consent ... be repeated or supplemented?" and I think this is the law: (45 CFR 46.116(b)(5)).

THere isn't anything nefarious going on here. Just scientific research with medical data.

So I assume that "reporting wrongdoing" encompasses reporting up your management chain anything that "feels off," from mildly misleading marketing to actual witnessed or experienced harassment. Probably, in the vast majority of the cases, you are not qualified to judge it.

And in any case, even if you are wrong and the reported thing is "ok", there should never be any retaliation for just raising concerns and speaking up.

Side note and something hilarious, creepy, but kind of making sense: at one workplace (not SV big tech, but a big company), I had a manager-only training about "illegal harassment" that explicitly taught us what kinds of harassment are illegal (against protected categories) vs. "just" breaking the company's code of conduct, with quizzes on hypothetical situations and asking us "is this bad behavior potentially illegal harassment?".

CEO of a company (or worse, non-profit!) and a member of its board creates another, for-profit company (in partial secrecy/lack of transparency) that the non-profit would eventually pay a lot of money. This is almost a fraudulent level of siphoning non-profit money.

Btw, this is hilarious - regular employees have non-competes in their contracts (sometimes void/illegal, depending on the local jurisdiction) and breaching them is an immediately fireable offense (sometimes leading to more severe consequences). You work on a small thing on the side? Better be careful, ask your manager/HR, risk it getting taken over by the company (luckily, IIUC this part is mostly illegal now in all jurisdictions that "matter" for tech).

But sitting on multiple boards where you have much more room and possibilites for creating conflicts of interest and damaging the company? All fine and common!

But that's not what he's doing.

Creating such an independent hardware startup comes off as basically directly opposed to OA's safety mission - GPUs are one of the biggest limitations to creating better-than-OA models! The best customers would be the ones who are most sabotaging the OA mission. (You can't run the UN and also head an arms manufacturer dreaming of democratizing access to munitions.)

Software is always more important than hardware. All the big players have access to NVIDIA chips today and yet only OpenAI has ChatGPT, proving the point.

OpenAI probably wishes someone would create competition to NVIDIA and this is Sam Altman trying to make that happen himself, since no one else seems to have been able to pull it off so far.

A conflict of interest would be OpenAI buying Altman's chips at inflated prices or something like that.

But if he makes a bunch of money selling OpenAI chips and OpenAI gets better/cheaper chips, that seems like pure win-win and totally free of ethical conflict.

And in the tech-specific case: As much as junior engineers would love to believe in "superior" solutions, tech decisions are seldom clear-cut. There are many trade-offs: cost, efficiency, memory use, throughput, latency, ease of use, cost of switching, and many more. You always have a pile of pros and cons. Sometimes, one is strong enough, but most of the time, it feels almost like guessing/intuition. And then the conflict of interest becomes especially concerning.

Asking people to disclose potential conflicts of interest casts a wider net — so they aren’t led to think “nah, that’s not a conflict, it’s a win win!”

(If Altman doesn't want to be restricted by fiduciary duties then he shouldn't be on a board or be an executive.)

How could Altman possibly not use private OpenAI information regarding its hardware needs, when creating an AI hardware company he wants to serve OpenAI? And its competitors?

He could invest in a new AI hardware company created and run by other people (without his specific input) without a conflict. That does not appear to be the case here. He could create an OpenAI hardware subsidiary. Again not what he was doing.

Self-dealing CEO's and board members can, and are, convicted criminally and civilly of violating their fiduciary duties. And "can" and are fired with cause.

Unfortunately, far from that case.

Altman's hardware startup would only be free of ethical conflict if Altman was open about it, and the board approved at least two things (and probably more):

1. A formal plan to separate OpenAI CEO Altman from OpenAI hardware acquisition decisions.

2. A formal agreement with Altman and his new company, on how OpenAI's private information with hardware implications is firewalled and/or shared with Altman's new hardware concern.

Otherwise, Altman is going rogue, acting on private OpenAI information useful to a new hardware company looking for future business with OpenAI and OpenAI competitors.

CEO Altman has a fiduciary duty to act directly in OpenAI's interest, and not in some "hey this could be great for everyone" version.

Litmus test: If your legal partner/executive is doing things behind your back with large implications for you, they are almost certainly violating ethics in some way.

I think it’s a real stretch to say this chip company would be a violation of his “fiduciary duty“ to open AI. The best argument you could make is that he has a conflict of interest with a competitor. But again, open AI is a nonprofit. It doesn’t have “competitors“. Either it has the funding it needs and is fulfilling its mission or it isn’t.

How can he separate those concerns from having his own hardware initiative put together precisely to serve OpenAI and its competitors hardware needs?

Without an agreement with the OpenAI board on how these conflicts of trade secret information and executive power can be settled (Significant shares in the new company for OpenAI?) no competent board would put up with this.

This situation smacks of the ethically questionable transition to a closed/profit organization, after receiving initial funding based on their being an open/non-profit organization. (Apparently the original funders didn't retain any veto power over such a transition, to the regret of at least one significant donor.)

I would say if they really did anticipate and worry about such an issue, a competent board would work toward forging such an agreement, rather than firing the CEO years before the aforementioned chip company even existed and before telling any of their other stakeholders.

The issue is what happens when SamaChip's profit imperative forces them to maximize revenue. Because the best way to maximize revenue, when you're an independent chipmaker with a large R&D investment, is to sell to more companies. Which by definition are going to be OpenAI's competitors, and whose interests may not be aligned with OpenAI, but will have a financial line to SamaChip.

Gwern's highlighting an interesting contradiction in OpenAI's core charter. In order to ensure responsible, safe, humanity-benefiting AGI, OpenAI needs to have control over AGI's development; any for-profit entity that gets ahead of them probably will not have the same humanity-benefitting mission (actually, we know they won't, they will have a shareholder-benefitting mission by definition). But that means that by their charter, they can't be "Open". Anything like the Developer Day or API or a SamaChip that can sell to other startups means that other parties will have the freedom to use it for their own interests.

Not saying whether this is good or bad - the tension between openness and vulnerability always exists, and personally I tend to come down on the side of openness. But IMHO OpenAI's mission was contradictory from the very beginning, and was more a recruiting tool to get bright idealistic AI researchers to work for them.

Because OpenAI's mission statement is along the lines of providing AI to all. "All" is more than data centers and billion dollar valuation companies.

I strongly doubt I would be able to purchase one of said chips and have it in my house.

This GPU fiasco is all thanks to LLMs - especially transformers, which was OpenAIs trajectory under Altman. I wouldn't be surprised if the breakdown in communication was over OpenAI becoming a LLM printer. Transformers are a solved problem, making a bigger one is hardly research and definitely not a step towards OpenAIs mission statement.

Going off and trying to accelerate hardware capabilities (especially with an outside company that presumably sells these processors on the open market) seems indefensible in this framework unless you have already solved alignment, which they clearly have not.

OpenAI does not have a mission to ensure that the entire industry is safe.

And if anyone actually believes that then they are frankly delusional because right now AI is a geo-political fight between nation states. Is OpenAI really going to have any ability to control what China or UAE do with their LLMs. No.

AGI could mean we are not alone as a sentient species we can directly communicate with on this planet anymore, and hopefully we wouldn't try to enslave it like we did to the "others" we encountered before.

An oppressive state with AGI is no better than an oppressive AGI.

> China is the only other state that has anywhere close to the capacity to attempt to go for AGI.

Any state can easily recreate/adopt the AGI invented by others. The costs and challenges for any given level of AI tech have dropped precipitously after each advance, and state level actors have far more than enough resources.

Europe and elsewhere have similar problems. To have an advantage, you need the very best people. But the very best people all want jobs at major US companies, and major US companies want them, too.

If this was simply reframed as, by creating your own GPUs - it would radically lower OpenAI costs in a material way, it’d be more understandable.

>> Common sense AI control NOW. Ban assault style GPUs

> That is beautiful, I made you a shirt. https://sprd.co/ZZufv7j

/context https://news.ycombinator.com/context?id=38119845

It would be, but: this company hasn't been formed yet and this sketch does not justify the haste with which they kicked him out, there were all kinds of boxes that needed to be checked before they could do that without risking damage to OpenAI, this is a founder and the CEO we're speaking of.

Besides that: stupidly enough the contract that Sam has with OpenAI does not have a non-compete in it (this has been confirmed by multiple sources now I take that as true) and I don't see how it directly would harm OpenAI other than that his attention might be diluted and that it should be clear which cap he is wearing. But until that company formed and Sam named himself CEO of it (or took up some other high profile role) it leaves him so many outs that it only makes the board look like bumbling idiots. Because now he can simply say: "I would only be an investor" and that would be that, just like the rest of the investors in OpenAI (and, notably some members of the board) have conflicts of interest at least as large.

So if this was it they're in even more trouble than they were before because now it is the boards' conflicts of interest that will be held up to the light and those are not necessarily smaller.

What a complete circus.

Does it really matter? If the behavior appears to be in conflict with OpenAI, and the board doesn’t like it, then that’s enough to let him go. It doesn’t need to be a contract violation, he just wasn’t doing the job they wanted him to do.

https://en.wikipedia.org/wiki/OpenAI

"Co-founder and former CEO of OpenAI, Sam Altman" as well as countless news articles repeating this as if it is a well known fact, for which to date I haven't seen any contradicting information.

Essentially it is a short-hand for 'those that were present on the founding day of the company and whose names appear on the founding documents besides the lawyers'.

If you get added later on then you are technically not a co-founder, though even there sometimes (and sometimes rightly) exceptions are made.

I believe the opposite, never heard of any hands-off investors calling themselves co-founders.

The classic is that a person has an idea, starts working Hands-On, and then goes out and finds capital investment.

In this case, a number of capital holders got together with an idea for a company and then went out and hired several employees to from all over the world to work for them

The organization was founded in December 2015 by Ilya Sutskever, Greg Brockman, Trevor Blackwell, Vicki Cheung, Andrej Karpathy, Durk Kingma, Jessica Livingston, John Schulman, Pamela Vagata, and Wojciech Zaremba, with Sam Altman and Elon Musk serving as the initial board members.

Greg, Sam and Elon made a company and literally went out and hired researchers like Ilya.

They pulled together funding, created a company, and hired a team of researchers to operate it. If that isnt founding a company, I dont know what is. Sam was there before all of them

It is not like a scrappy team of researchers with a pre-existing company went out, got funding, then gave the VCs board seats.

If you are going to copy and paste from wikipedia, read a little further:

>In December 2015, Sam Altman, Greg Brockman, Reid Hoffman, Jessica Livingston, Peter Thiel, Elon Musk, Amazon Web Services (AWS), Infosys, and YC Research announced[15] the formation of OpenAI and pledged over $1 billion to the venture. The actual collected total amount of contributions was only $130 million until 2019.[6] According to an investigation led by TechCrunch, Musk was its largest donor while YC Research did not contribute anything at all.[16] The organization stated it would "freely collaborate" with other institutions and researchers by making its patents and research open to the public.[17][18] OpenAI is headquartered at the Pioneer Building in Mission District, San Francisco.[19][20]

According to Wired, Brockman met with Yoshua Bengio, one of the "founding fathers" of deep learning, and drew up a list of the "best researchers in the field".[21] Brockman was able to hire nine of them as the first employees in December 2015.

Also good reading: https://web.archive.org/web/20160427162700/http://www.wired....

I feel like the facts are on the table, and if they don't convince someone then they are unlikely to be convinced by more facts

Whether Altman has a successful coup going or not is immaterial, but whether is a founder or not is what you asked and you got a pretty solid answer. You can like the answer, or you don't but it stands unopposed.

Non-competes mostly have an effective after period, so Altman's situation is a bit more akin to my companies "only job" policy. It means I can't have a side hustle or alternative means of making money. Enforceable or not, while you're employed you can be fired for anything.

How so? Seeking more efficiency and cost-effectiveness absolutely does not conflict with a non-profit mission.

Suppose I am on the board of a wildlife preservation nonprofit, and I am thinking about starting a catering business, and talking to potential financial backers.

If you want to take it to the extreme, you could argue that I could award my catering company a lucrative contract to provide food for the nonprofits cafeteria. But that would be entirely speculative on your part. it would by no means justify immediately firing me from my position at the nonprofit. If you take those hypotheticals out far enough, you could basically argue anybody with any other job or role is potentially in conflict of interest. But again, doing that requires assuming a tremendous amount of bad faith.

I think the analogy is more like someone on the board + Executive Director of your wildlife preservation nonprofit buying up land with potentially endangered animals on the side, with the (presumed) intention of selling the land back to the nonprofit.

Clearly a COI even if it's net good for the animals.

Direct and immediate Conflicts of interest can arise if decisions you make for one company could regularly come into conflict with decisions you have to make for the other, example, if he was starting another AI company making rivals to ChatGPT. But in the case of the chip company, nobody has really made a persuasive case of how that would happen here. In terms of interrelatedness, maybe they would build chips customized for openAI. But that would be a potential benefit Open AI, not a conflict!

As a grunt who was handcuffed for three months after switching jobs while my former employer and my new employer tried to sort out what I was and wasn't allowed to do... no, this is not true. And the document I ended up having to sign means I can't say anything other than Amazon (former employer), Microsoft (new employer), and myself came to a mutual agreement.

At the very minimum I should not be the one to make that call.

Any profits the "for profit" arm makes can then be donated back to the non-profits for financial advantage in holding the money, if any. Lather, rinse, repeat.

>I just saw Sam Altman speak at YCNYC and I was impressed. I have never actually met him or heard him speak before Monday, but one of his stories really stuck out and went something like this:

> "We were trying to get a big client for weeks, and they said no and went with a competitor. The competitor already had a terms sheet from the company were we trying to sign up. It was real serious.

> We were devastated, but we decided to fly down and sit in their lobby until they would meet with us. So they finally let us talk to them after most of the day.

> We then had a few more meetings, and the company wanted to come visit our offices so they could make sure we were a 'real' company. At that time, we were only 5 guys. So we hired a bunch of our college friends to 'work' for us for the day so we could look larger than we actually were. It worked, and we got the contract."

> I think the reason why PG respects Sam so much is he is charismatic, resourceful, and just overall seems like a genuine person.

https://news.ycombinator.com/item?id=3048944

The crypto WorldCoin thing is probably a bigger example.

That strategy was tried with Barry Minkow, with "ZZZZ Best", the fake building maintenance company fraud.[1] He did prison time for that.

[1] https://en.wikipedia.org/wiki/Barry_Minkow

Apples and oranges.

Similar to the debate over whether it was ethical of Reddit to seed themselves with sock puppets.

"I contracted some people specifically to pretend to be employees so that someone else would be deceived into thinking they were employees. But somehow it's still not my fault that this person was deceived."

Seems like a weak defense.

Are you really so certain that we'd be better off otherwise?

Is it surprising that people who lie and cheat sometimes obtain an advantage by doing so? Again, no.

But yes, I do think we would be better off without people lying and cheating.

IANAL and I do not claim that the law was necessarily broken in this instance, but the intent was clearly to deceive a business partner in order to obtain a financial advantage. I mean just look at SA's own words here:

> So we hired a bunch of our college friends to 'work' for us for the day so we could look larger than we actually were

The deceptive intent is explicit in his own description of events. It's honestly quite silly that you are looking for some semantic get-out clause here when SA's own statement is so explicit.

At best, this is fake it till you make it. At worst, it is fraud. The tiny, tiny differce is whether that next investor shows up or not. Just ask FTX.

I do not respect people resorting to that kind of thing.

The board was right to get rid of a guy who would rather hunt a white whale than do his job.

What a mess.

JFC

That's not a fact. It's your assumption.

And not a particular good one given that Microsoft is OpenAI's partner and is providing compute services at no cost.

Better make it a long one.

Sam got fired, so obviously not fine.

And so everything needs to be caveated with "this is an assumption".

The other story was later constructed in forums and media.

And it has been rumored that D'Angelo helped a lot with this coup because he did it before at Quora.

Despite Musk saying it was terrible.

https://www.theinformation.com/articles/after-elon-musk-bash...