DebtDeflation

Circumvents regulatory oversight and also shafts 99% of the employees. Seems to be a backdoor way to acquire the key founders/leaders and IP (via a perpetual license) while leaving behind a desiccated husk of rank and file employees, customers, and obligations.

Click through to the law in question. It's the Civil Code not Criminal Code, and states, "an individual who has been harmed by a violation of this title, including, but not limited to, unauthorized access or use of ALPR information or a breach of security of an ALPR system, may bring a civil action in any court of competent jurisdiction against a person who knowingly caused the harm."

So you have to prove actual harm. You have to identify the individual person who caused the harm. You have to prove they knowingly caused the harm. You have to quantify the harm in monetary terms. Then you can sue them for actual damages + attorneys' fees.

>what is dark matter?

I suspect in the end it will turn out to neither be exotic new particles nor modifications to gravity, but rather that there is something fundamental about large scale structure formation in the universe that we just do not understand at the present.

I've been reading commentary on aviation forums like PPRuNe and the consensus amongst pilots and others in the field seems to be that "electrical glitch" is a near impossibility. Also, the report states "and immediately thereafter, the Engine 1 and Engine 2 fuel cutoff switches transitioned from RUN to CUTOFF position one after another with a time gap of 01 sec". They wouldn't have said the switches transitioned from one position to another if they weren't certain, they would have said something like a voltage change consistent with cutoff occurred. Several people have pointed out that these are large mechanical switches that make a distinctive noise when moved and the CVR is sensitive enough that it would have picked it up. Also, it's important to note that no safety bulletins were issued to Boeing as they would have in the case of some electronics issue.

Now, when I say deliberate act, I don't necessarily mean "pilot suicide" though that is certainly a strong possibility.

I'd say the odds are 50%. The odds of the opposite scenario - where the pilot who said "did you flip the fuel cutoff" wasn't the one who did it are also 50%.

Based on the cutoffs for both engines being flipped 1 second apart, the above exchange being caught on the CVR, and then within 10 seconds the (presumably the other) pilot switching them back to Run, it's pretty clear that this was a deliberate act.

Hold up. GenX'er here, graduated college in the mid 90s. Are you telling me that college keg parties in the basements of off-campus housing is no longer a thing?

It's true that "AI detection algorithms" are not particularly reliable.

It's also true that if you have fake CITATIONS in your works that such algorithms aren't necessary to know the work is trash - either it was written by AI or you knowingly faked your research and it doesn't really matter which.

It's not just about smaller models. I recently bought a Macbook M4 Max with 128GB RAM. You can run surprisingly large models locally with unified memory (albeit somewhat slowly). And now AMD has brought that capability to the X86 world with Strix. But I agree that how long Google, Meta, Alibaba, etc. will continue to release open weight models is a big question. It's obviously just a catch-up strategy aimed at the moats of OpenAI and Anthropic, once they catch up the incentive disappears.

Oh 100% the comparison to dotcom is better than crypto. Despite the hype, AI and dotcom both were useful, crypto was always a grift.

If anything, the LLM overhype is starting to die down........to make way for the AI Agent hype which is on trajectory to be 1000X worse. People are writing articles and making videos about how AI Agents will replace SaaS. What?

Pure propaganda. The people who were laid off were in the US, Canada, and Western Europe. The people who were "rehired" were in India. AI had a minimal role, primarily as a cover story.

Watch them cancel 18A and claim it's because 14A is coming along faster and better than planned. Like they did with 20A.

You don't even need AI. Just a bot that waits until the end of the meeting and then says, "Nothing from me. Thanks everyone."

How does that even work? You take away thousands of people in the market selling (presumably these were enterprise sales, not low end consumer stuff that "sells itself") and what happens to revenue? This strikes me as classic bean-counter logic that sees all sales and marketing effort as pure cost and assumes it can be eliminated while simultaneously viewing revenue as a constant, with no relationship between the two. Really an extension of the view that you should just eliminate all of the engineers first, because they're the most expensive. Too many people in the C-suite detached from their products and their customers, who see their company as just numbers in a spreadsheet.

>how do you have a layoff every couple of months for 3 years

It has absolutely nothing to do with managing out low performers or managing existential business risk. It has everything to do with managing annual EPS to Wall Street's expectations. There was an inflection point at the end of 2022 where Revenue growth slowed so to maintain Earnings growth, costs had to be cut continuously, and this process is still playing out.

It's wild to me how this movement wants to bring back all of the lowest paid, least value add, lowest skill jobs back but is totally ok with shipping highly paid, highest value add, highest skill jobs off to India. I understand that it's all a grift, the marks only care about the former jobs, and those jobs are never actually coming back, but still.

This is the most succinct description of our current reality that I've ever seen. Kudos!

And in the meantime, their goal is clearly to make money off non-AGI AI.

I constantly get quasi-religious vibes from the current AI "leaders" (Altman, Amodei, and quite a few of the people who have left both companies to start their own). I never got those sort of vibes from Hinton, LeCun, or Bengio. The latest crop really does seem to believe that they're building some sort of "god" and that their god getting built first before one of their competitors builds a false god is paramount (in the literal meaning of the term) for the future of the human race.

These examples of double standards for labor vs capital are literally everywhere.

Capital is supposed to be mobile. Economic theory is based on the idea that capital should flow to its best use (e.g., investors should withdraw it from companies that aren't generating sufficient returns and provide it to those who are) including being able to flow across international borders. Labor is restricted from flowing across international boundaries by law and even job hopping within a country is frowned upon by society.

We have lower rates of taxation on capital (capital gains and dividends) than on labor income because we want to encourage investment. We're told that economic growth depends on it. But doesn't economic growth also depend on people working and shouldn't we encourage that as well?

There's an entire industry dedicated to tracking investment yields for capital and we encourage the free flow of this information "so that people can make informed investing decisions". Yet talking about salaries with co-workers is taboo for some reason.

The list goes on and on and on.

Likewise, we've been building conversational interfaces for stuff like this for over a decade using traditional NLP techniques and orchestration rather than LLMs. Intent classification, named entity extraction, slot filling, and API calling.

Processing returns is pretty standardized - identify the order and the item within it being returned, capture the reason for the return, check eligibility, capture whether they want a refund or a new item, and execute either. When you have a fully deterministic workflow with 5-6 steps, maybe 1 or 2 if-then branches, and then a single action at the end, I don't see the value of running an LLM in a loop, burning a crazy amount of tokens, and hoping it works at least 80% of the time when there are far simpler and cheaper ways of doing it that will work almost 100% of the time.

Meanwhile, good old aspirin is not only an effective pain reliever but it has also been shown to reverse the effect of cartilage loss in osteoarthritis:

https://www.sciencedirect.com/science/article/abs/pii/S03785...

And that's not even taking into account its cardioprotective effects.

But hey, it can be hard on the stomach in SOME people so we developed alternatives with all sorts of nasty side effects, up to and including heart attacks and strokes (rofecoxib, valdecoxib, etc.) Hell, a decade ago, the FDA issued an advisory across the board that "Non-aspirin NSAIDS increase the risk of heart attacks and strokes":

https://www.fda.gov/drugs/drug-safety-and-availability/fda-d...

Some day we will look back on the attempted replacement of aspirin with these other drugs the same way we now look back on the replacement of butter with margarine (and the replacement of saturated fats with trans fats more broadly).

Almost every enterprise use case is a clear workflow use case EXCEPT coding/debugging and research (e.g., iterative web search and report compilation). I saw a YT video the other day of someone building an AI Agent to take online orders for guitars - query the catalog and present options to the user, take the configured order from the user, check the inventory system to make sure it's available, and then place the order in the order system. There was absolutely no reason to build this as an Agent burning an ungodly number of tokens with verbose prompts running in a loop only to have it generate a JSON using the exact format specified to place the final order when the same thing could have been done with a few dozen lines of code making a few API calls. If you wanted to do it with a conversational/chat interface, that could easily be done with an intent classifier, slot filling, and orchestration.

I didn't mean to imply he started it alone. Though his co-founder Lucy Guo is almost as bizarre of a story as Wang himself. I'm curious, what were they doing before data labeling?

It's a wild story for sure. Dropped out of MIT after freshman year and starts Scale to do data labeling. Three years later Scale has a $1B valuation and two years after that Wang is the world's youngest billionaire. Nine years after Scale's founding they're still doing less than $1B in annual revenue. Yet Meta is doing a $14B acquihire. There's definitely more than meets the eye. I suspect it involves multiple world governments including the US.

This is exactly it. QC right now is a series of very cool science experiments that are being marketed to [Government Officials, CEOs, Investors, the General Public] as product development, which it is not. We're at the stage of scientists in the 1910s creating the first vacuum tubes and noting the ability to amplify and control small currents with larger ones but these companies are pretending it's instead the early 1980s with the PC and 8088 and Moore's Law getting ready to take off like a rocket.

My assumption when the story broke was that the 700 engineers were using various AI tools (Replit, Cursor, ChatGPT, etc.) to create code and documentation and then stitching it all together somewhat manually. Sort of like that original Devin demo where AI was being used at each step but there was a ton of manual intervention along the way and the final video was edited to make it seem as if the whole thing ran end to end fully automated all from the initial prompt.

>Agent is clear in that it acts on behalf of the user.

Yes, but you could say that AI orchestrated workflows are also acting on behalf of the user and the "Agentic AI" people seem to be going to great lengths to distinguish AI Agents from AI Workflows. Really, the only things that distinguish the AI Agent is the "running the LLM in a loop" + the LLM creating structured output.

Envelope #2 in a game of 3 Envelopes.

> IBM’s error correction algorithm, known as a low-density parity check code, will make it possible to use 12 physical qubits per logical qubit

They have a machine with over 1000 PHYSICAL qubits today. I'm guessing this 12 physical per one logical is some sort of theoretical possibility, not something that can actually be realized.

Then you probably need a few thousand LOGICAL qubits to do anything meaningful with a QC.

There is no evidence that any of this will scale.

Every news release seems optimized for generating hype that will produce a round of funding that will enable the next news release.

As of the early 2000s, ATC was still using vacuum tubes. In fact, the FAA was the single biggest buyer of vacuum tubes in the world at the time, almost all of them sourced from former Soviet bloc countries. I think they've all been replaced by now, but I can't say that with 100% certainty.