I worked in a job a decade back where I was the only tech guy and had a special 128 GB RAM machine. All the 'Big data' for the team was done by me using R tidyverse, data.table and few libraries and they thought of it as magic as there were few tech people there.
Still feel a lot of enterprises and industries looked over its capabilities then.
With LLMs the challenge of R syntax is a little easier for data analysts to climb, especially the new ones.
there is enough proof that they had a chatbot internally which was quite competitive but was not pushed through for all these fears, it seems they were always confident that they could catch up and scaling laws were their internal defense.
The question now though is neither might have expected Chinese labs to catch up so fast.
They did, but the goalposts keep moving, so to speak. We're approximately here : advanced semiconductors, artificial intelligence, reusable rockets, quantum computing, etc. Chinese will never catch up. /s
Yet tbh if the US industry had not moved ahead and created the race with FOMO it would not had been easier for Chinese strategy to work either.
The nature of the race may change as yet though, and I am unsure if the devil is in the details, as in very specific edge cases that will work only with frontier models ?
That is amazing if they can do all of this at < 10 % of the cost of frontier labs. Off course they work in the shadows of the great work done in the frontier labs and shared, but there is some exceptional high speed execution happening behind the scenes that shows this is clearly a race, but a race where China is happy to be #2 as long as the gap is not significant and the costs are reasonable
Frankly, I am pleasantly surprised to see that being a relatively close number two seems to be both practical and is turning out to be enormously beneficial to humanity. I am concerned that deep secrecy on OAIs part could change that, but it’s also possible that the genie is sufficiently out of the bottle that it no longer would be practical.
Fair, but the 75% margins can be reduced to 25% with healthy competition. The lack of competition in the frontier chips space was always the bottleneck to commoditization of computation, if such a thing is even possible
Sometimes I wonder who the rational individuals at the other end of these deals are and what makes them so confident. I always assume they have something that general public cannot deduce from public statements
Most of the money flowing to the big players is from tech giant capex, originally from net cash flow and lately its financed by debt. A lot of these investors seem to now essentially be making the case that AI is "too big to fail". This doesn't at all resemble VC firms taking a lot of small bets across a sector.
If the whole market goes to bet at the roulette, you go bet as well.
Best case scenario you win. Worst case scenario you’re no worse off than anyone else.
From that perspective I think it makes sense.
The issue is that investment is still chasing the oversized returns of the startup economy during ZIRP, all while the real world is coasting off what’s been built already.
There will be one day where all the real stuff starts crumbling at which point it will become rational to invest in real-world things again instead of speculation.
(writing this while playing at the roulette in a casino. Best case I get the entertainment value of winning and some money on the side, worst case my initial bet wouldn’t make a difference in my life at all. Investors are the same, but they’re playing with billions instead of hundreds)
There isn't necessarily rationality behind venture deals; its just a numbers game combined with the rising tide of the sector. These firms are not Berkshire. If the tide stops rising, some of the companies they invested in might actually be ok, but the venture boat sinks; the math of throwing millions at everyone hoping for one to 200x on exit does not work if the rising tide stops.
They'll say things like "we invest in people", which is true to some degree, being able to read people is roughly the only skill VCs actually need. You could probably put Sam Altman in any company on the planet and he'd grow the crap out of that company. But A16z would not give him ten billion to go grow Pepsi. This is the revealed preference intrinsic to venture; they'll say its about the people, but their choices are utterly predominated by the sector, because the sector is the predominate driver of the multiples.
"Not investing" is not an option for capital firms. Their limited partners gave them money and expect super-market returns. To those ends, there is no rationality to be found; there's just doing the best you can of a bad market. AI infrastructure investments have represented like half of all US GDP growth this year.
Have most Western countries mostly given up on supply side policies for housing price stablisation as an option ? Odd given how much construction tech has improved.
The walls of a house are already cheap to build with a low skill barrier etc. Land, foundation, windows, doors, roof, plumbing, appliances, yadda yadda add up, but everyone seems to focus on walls.
If you really want to drive down the cost of housing figure out cheap windows or start convincing people they don’t need nearly as many of them.
Don’t believe me? Compare the cost of a large 2 story prefab shed ~10-20$/sf without amenities like plumbing vs a house of similar size.
Even that could run you near 2,000$ on a modest new home ignoring the rebate.
But my point was more that price savings are easy to slip into the new construction pipeline for a wide range of homes. Similarly Argon helps r with insulation on day 1, but only lasts ~20 years on a window that may be in use for 100.
Housing prices won't ever stabilize as long as land is a tradeable commodity that 'stores' value. If people can make money with man-in-the-middle techniques and cheap money (low interest rates), then they will, and this will drive the real-estate boom-bust cycle we have that runs on ~18 year cycle (last bust was late 2008, early 2009, prior bust was around 1990)
Become communist (or communist lite) and reserve land ownership to the state, lock land price to its productive capacity (as assessed by the state), mediate sales and usage taxes based on the same. Prevent resale/sublet between individuals. Allow transfer of control only from individual -> state -> individual.
Or use a regulatory approach where only the state can set the price of land, and the price is based on well-defined factory (productivity, proximity to population centers, population density, etc).
All value in an economy arises from real estate. Real estate (and the related businesses, including banking and lending) drive all economic activity in the world.
Some cities in Europe have famously tried to place a ceiling on the price of a square meter, to varying degrees of success. The German federal government has notoriously reversed Berlin's decision to do so.
Since land belongs to everybody and to no one, perhaps we should let the government decide what to do with it. Or at least promote more co-op housing and land trusts.
Just like the example of US healthcare yesterday where someone successfully negotiated cash rate of 194k to 33k I do not think it will be scaleable as hospitals will push back with new regulations or rules.
Is it? Usually the professional tools are going to be incredibly more powerful and precise than the consumer grade stuff. That would be true here just as much as with previous iterations of sales. The opposing side has an information advantage and could expose their knowledge of true prices in the form of some RAG dataset, while the consumer grade LLM would just have to guesstimate. The information disadvantage doesn't disappear because it's machines doing the negotiating.
In addition, consider that one could train a professional-grade sales LLM against all the available "general purpose consumer" models with adversarial training techniques, so that it can "beat" them at price negotiation. Just as a quick sketch, you could probably do some form of prompt injection to figure out which model you are talking to and then choose the set of tokens most likely to lead to the outcomes you want.
Finally, the above paragraph assumes that such a sales LLM couldn't just buy certain responses from the consumer grade LLM provider btw, similar to how you can buy ad space from Meta and Google today.
also because it is modular which really works for the Global south, it can be taken to demand centers and demand adjusted to the supply to a small extent (e.g. irrigation pumps)
Still feel a lot of enterprises and industries looked over its capabilities then.
With LLMs the challenge of R syntax is a little easier for data analysts to climb, especially the new ones.
reply