Unlike the 90s and dotcom companies, there are few if any tiny AI stocks that offer huge returns. All the companies are private, or you have to invest in huge companies like Microsoft or Google. If anything, this protects smaller investors. Open AI is not traded on any exchange.
Yes, this isn’t news, it’s speculation about what might happen. Startups stay private longer, too. So conditions don’t seem ripe yet. But it seems plausible that AI startups will eventually go public?
Generative AI is very useful for creating rigged demos, which are particularly useful when people want to believe. Widespread skepticism might have some effect, but there are always skeptics, and no particular reason to believe there won’t be any winners to make venture capitalists happy with a few of their bets, enough to pay for the failures.
The only access investors have to outsized gains are through investing in the companies that invest in these companies. What small investors need are private equity index funds.
Someone else mentions Netflix vs. Blockbuster. Small investors would have been doubly screwed if they could only invest in Blockbuster because Netflix was still private.
>Unlike the 90s and dotcom companies, there are few if any tiny AI stocks that offer huge returns.
The number of (legitimate) publicly-traded internet-related companies in the late 90's that were accessible to the average investor was also incredibly small. Legitimate small companies are rarely traded on exchanges in general, as they have no reason to solicit investment from the general public. It's the same phenomenon behind crowdfunding; if you have a legitimate product in development, you can simply convince a conventional lender to loan you the initial capital.
In this context I see crowdfunding as a cheap way of gathering data of breadth of interest in your product that can be taken to a conventional lender. Otherwise I completely agree.
Hey, this is what I appreciate about US investors, they rather put their money in BS moon-shot ideas that will 99% fail, but this way they fuel a high demand for SW jobs leading to competition and high wages in the industry, instead of putting all their money in stuff with a 99% success rate like real estate, like investors do here in Europe.
The fact that investors are constantly chasing new and risky ideas, instead of just zero sum assets, is why the US has the best jobs market in the world.
If investors and consumers were responsible and conservative with their money, dev wages would be much much lower. So enjoy it, because you have it good.
This is right. Try stuff, invest energy in building things, no one knows in advance what will take off. It is not bad or wasteful in any real sense. Money is only useful when it's motivating action. If an idea is bad that branch of investment will eventually wither and die if nothing useful is found. You do not need to take personal offense if you think a current trend is dumb. You might be wrong, you might be right, the only thing I can say for sure is that whining about it is lame, and if you think you know better then go and try to build something else and make bets on the trends you think will become real waves.
Old curmudgeon that I am, I often suggest to people who are sure they know better how markets will move that they invest appropriately (you can even make investments that payoff if the market goes down!).
Of course, none of them ever do.
Investing your own money is a heluva lot different, and it's where the rubber meets the road.
The thing is, without those risky hypergrowth BS companies, your jobs market for "safe jobs" would be shit. So you do have both options available at the same time.
Nobody is forcing you to work at FAANGs or other hypergrowth companies that have redundancies, there's older established companies that didn't have redundancies which of course, pay much less than FAANGs do.
Places that only have safe employment options, without hypergrowth, also have massively lower wages for everyone.
It's not about "safe jobs", it's about fostering diverse commercial environments in which people can readily get good-fit "new jobs" when their old job disappears. This sort of environment has the side-effect of increasing wages for everyone.
When market contractions hit even "safe jobs" aren't necessarily safe.
Competition is important. But there are limits, after which it's better for everyone if the business-creator comes up with a novel idea and pursues that instead of iterating on everyone else. This is very difficult. It might be a bit less difficult if the culture didn't encourage followership in its leaders and funders.
US investors are not doing this out of generosity. European elites built up their wealth with land holdings and real estate over centuries. In most cases, the value of these assets are unknown. There is little benefit to be gained from disposing of that wealth to speculate on the latest fad and facing taxes.
The Duke of Westminster for example was the youngest billionaire in the UK with an estimated net worth of £10bn at 35. His family's land includes 100s of acres of the most valuable real estate in London. Their ancestor was granted these lands by William the Conqueror in 1066.
>US investors are not doing this out of generosity.
Of course, neither are European investors, but that's not the point. The point is a dinamic jobs market where the money goes on ideas and businesses, insted of assets, pushes all wages up and benefits the middle class.
Investors living off land and other such rentseeking endeavors destroys the middle class as it's just feudalism with extra steps.
>There is little benefit to be gained from disposing of that wealth to speculate on the latest fad and facing taxes.
Of course, but this is the fault of our tax system which incentivizes land hoarding and rent seeking by those already afluent, at the expense of the have-nots looking to buy in.
I let out a family house. The rent is the greater part of my pension. I used to live there, until I was divorced (it was too big for one person). It's been clear to me for decades that freehold land was a better investment (here) than investing in a conventional pension; and I lived through the outright theft by Robert Maxwell of the whole of his staff's occupational pension fund. So I paid off my mortgage as quickly as I could.
Regarding the chores, that someone mentioned upthread: I have a long-term tenant, a family. Their kids go to school around the corner. They mow the lawn. They always pay the rent by bank transfer. I hire tradesmen to deal with the gas, plumbing and electricals. I visit twice a year to check that everything's OK. These aren't really problems.
And rental housing is scarce where I live. It's a university town with several important hospitals, and surrounded by a no-development green belt. I'm providing a social benefit.
I'm not sure when this kind of "rent-seeking" turns into the kind of rent-seeking of which people disapprove.
>I'm not sure when this kind of "rent-seeking" turns into the kind of rent-seeking of which people disapprove.
Where I live there are individuals owning entire apartment buildings, >10 houses, in fact, 80% of the entire country's wealth is owned by the top 300 richest individuals.
Single individuals renting out their spare home, doesn't even register on the radar compared to what the rentseeking of the truly wealthy looks like.
Rent seeking is a term coined by economists. It has absolutely nothing to do with what you are doing as a landlord (providing housing for money).
An example of rent seeking would be if you lobbied the local council to restrict new building or additional rental units in the area leading to artificially higher rents. It has nothing, directly, to do with receiving income for renting a housing unit
> ...but this way they fuel a high demand for SW jobs leading to competition and high wages in the industry,
There are many sides to this. Over a decade ago, Apple colluded with other tech companies to artificially suppress wages by crafting "no-poaching" agreements. Facebook refused to play ball.
99% success rate in real estate? I've lost money on real estate several times. I avoid it now. Most people seem to think that if they bought a house for $100,000, and sold it for $200,000, they made a $100,000 profit.
Nope.
There's:
1. property tax
2. insurance
3. maintenance
4. upgrades
5. dealing with the lawn
6. 6% commission taken by the real estate agent
7. inflation
8. all the time spent dealing with it
9. mortgage interest
10. the costs from having it sit empty on the market waiting for a buyer
11. and the most neglected cost, the time value of money
#1, #2, #3, #4, #5 are all costs of use of the property, not ownership of it. Either you're using it for yourself (In which case, you'd either be paying yourself for these things, or paying your landlord), or you're letting it out to someone (In which case, they are paying for use of the property).
If you just let the property rot without a tenant, then, yeah, this punishes you. I'd recommend against letting a property rot without a tenant, just like I'd recommend against buying 500 barrels of crude oil, and letting them rot away into worthless sludge in your cellar over twenty years. All things that we buy have a limited shelf life, you need to use and maintain them before they expire. Structures built on a property are no different.
#6-11 are real costs of investing into property, and I agree - they cost a hell of a lot more time and money than buying an index fund and sleeping on it for 40 years.
Not really, there's a real cost of ownership for real estate that is more expensive than for owning stock in a company for instance. You still need a place to live and owning can be great for that – as to being a landlord there might still be better ways to make a return.
But the real problem with real estate "investment" is that the return doesn't come from doing something productive but rather is predicated on preventing someone else from doing something productive, namely making more buildings.
What really helped the "natural stupidity" is that paradoxically it actually worked better.
The really big winners were those which were undervalued because nobody expected them to grow as much as it would. Back in the 90s: "A mail order rental service growing bigger than Blockbuster? Get real." Being rational would filter the successful outliers along with the vast field of bad ideas. Usually going unfiltered is a losing proposition like playing the lottery. However, if the wins are outsized enough it perversely "pays to be stupid" assuming you have enough capital to win once and of course that similar market conditions emerge again.
That said useful applications of AI which may be implemented relatively easily are clear big winners as investments but that is pretty tautological and unhelpful for actually figuring out if say, an "AI love letter writing service" would be a terrible idea or not.
To me the lesson isn't to completely give up on trying and fund literally anything, rather it's that while some ideas seem worse, are worse... they are not that much worse. They are not that much worse that there isn't a price at which they are worth investing in.
That's less of "stupid" and more of "lucky to hit the niche at the right moment".
For every company that hit the niche and made billions over it there is few that were too early, and few that were not lucky enough to get the opportunity that made the winner win.
OpenAI out here selling pickaxes. Doesn't matter if half of people's ideas are trash and investors get fleeced, they get paid for api access all the same.
The success of ChatGPT (3.5 and 4 alike) is happening almost on as many dimensions as its latent space has.
OpenAI is selling pickaxes. It's also selling fear in every color. Fear for your future. Fear for humanity's future. Fear for the future. It's selling hope - of new profits, new discoveries, new futures. It's selling prophesy and it's selling religion.
It's also selling entertainment. To be honest, I bet that majority of use of ChatGPT still classifies more as fun than work. I pay for GPT-4 twice (ChatGPT+ and pay-as-you-go API access), and I use it daily, and 90% of my use is... exploratory in nature. I still consider it money well-spent. The problems may be real, and the outputs may be useful, but the choice of using GPT-4 is still dictated mostly by... playful curiosity.
(Also, it's the first tech thing in almost a decade that broke through my layers of cynicism and got me excited again. There's lot of cool things to play with, lot of discoveries to be made, and the progress is real.)
Whether or not LLMs are intelligent, OpenAI won. Financially, for now, but they'll have a spot in world history (for as long as history is a thing). The genuine technological jump they introduced... I can't remember any tech company in my life making this big a change in one step.
A lot of what those investors do (especially VCs on commission and VPs in big corps) is no different than gambling: they look at the founders, listen to their pitch and make a bet. Once big investors realise that GPT4 makes better bets, all those gamblers will have to follow that green Exit sign.
A process that leads to more of people’s wants being fulfilled. By this definition, there is maybe mild productivity benefits from gambling, but nothing compared to giving money to productive enterprise that is resource constrained.
Many enterprises "produce" objectively horrifically negative value for humans. It's not a rule that putting money in a gambling machine produces worse outcomes than if it was placed into a company.
If we view generally human lives today as better in the past (which I do), then it seems like it follows that the average capital investment is going to net valuable enterprise?
The only difference is a special exemption in the law for investing, as otherwise it would fall under the definition of gambling. Investing is gambling for big guys.
That’s just wrong. The comment you replied to is correct. In gambling the odds are against you. In investing the odds are with you (if they’re not, you’re doing it wrong!) you can get lucky or unlucky in either, but one has a negative expected return over time while the other has a positive expected return.
GPT is not going to make good bets. It has no current knowledge of the world, no logical thinking ability. Some investors don’t either, so the comparison might not be very unfavorable, but I don’t think Wall Street needs to be watching over their shoulder for GPT.
I've seen many "business" journalists complain bitterly about Amazon, Microsoft, etc., making lots of money. I suspect the bitterness comes from them having not bought those stocks early on.
All FAANG companies had dedicated money pools for share buybacks. Apple even pays dividends. The common belief is that is a company is doing share buybacks, it doesn't innovate anymore. These buybacks have been taking place for the past 10 years and they are still growing.
They’re one of the most tax advantaged way to return money to shareholders. That’s it. I don’t like this method or the need, but the motivation isn’t rooted in lack of innovation etc. Friedman doctorine has become gospel, so this is the result.
The charge of Lack of innovation is more that they returning money to shareholders instead of funding more r&d (organic/acquisition) not how (dividends/buybacks )
Some level of R&D is often needed to sustain a company's viability long-term. However, as a shareholder I also expect some share of profits to be returned to me.
Invest in upstream/ downstream businesses or in capital heavy projects, You could acquire also grow inorganically. R&D in the accounting sense is very different from research as we think. Yes there is diminishing returns for spending cash but I don't think these big companies are anywhere close to that being a real problem.
In a sibling post I talked about Google's inability to spend, this is also true for old school Blue chip companies like say Oil& Gas, they have had enormous record profits last couple of years and while they have for good reasons not doing lot of new wells[1] they have not not also done much green energy investments either.
[1] Financially sound not do so, given the long term prospects of oil are not great, and short term price fluctuations are not good decision basis for 30+ year projects new drilling entails.
Presumably, absent tax advantages of buybacks, more companies would pay dividends and bigger dividends. (At the end of the day a, simplistic, way to look at the innate value of a company's stock is the net present value of its future dividend stream.)
There is an additional difference beyond the tax advantage. With dividends, the market has more expectation of maintaining similar level of dividends, whereas there's generally and currently no similar expectation on companies doing the share buyback to buy similar amounts quarterly. This allows a bit more flexibility on the company. This extra flexibility may or may not be a good thing, but regardless, it is an additional meaningful difference.
Fair. And, historically, companies known for being consistent dividend stocks got hammered if they cut back. Which, as you suggest, may or may not have been a good thing in general. Whereas buybacks haven't really acquired the same level of X% per quarter. (And, in fairness, is inevitably caught up with questions of whether the stock is "fairly" priced.)
It's a way to dodge dividend taxes. It can be also be used to do some light price manipulation to boost executive compensation tied to share price. I don't think it's a huge evil or anything, but that would be the case for it being wrong.
The result of buybacks is that the number of shares drops. E.g. Berkshire's stake in Apple raised from 5.2 to 5.5 % merely as a result of Apple's buybacks.
Obviously shareholders benefit from having a larger piece of the cake while not requiring to pay any tax on this operations.
If I bought a stock yesterday and they issued a dividend today, there is no way for me to avoid the higher tax rate.
If I bought it yesterday and they decide to do a buyback today, I can hold it until the lower capital gains rate kicks in, thereby avoiding the higher tax rate.
See how I can avoid paying the higher tax rate?
Play some more games with tax loss harvesting, and the buyback option gives you a lot more ability to pay lower taxes. With the dividend there is a limit to the amount of loss you can offSet with short term losses.
Simply put, in any given situation capital gains is likely to give you a better option to pay lower taxes.
The practice itself is whatever, but the issue is that dividends are taxed differently than capital gains. At the end of the day the investor makes the same pre-tax income with either strategy, but they can face drastically lower tax burdens.
In other words, buybacks are a way of paying way less tax.
You can decide to hold long enough for the rate to go from short term to long term capital gains, which is something you can’t do with an unqualified dividend. There is also a limit to the amount of unqualified dividends you can offset with losses, which isn’t true for capital gains.
It is absolutely different because in order to recoup the benefit of a stock buyback you need to sell your holdings. Dividends, by contrast, accrue to those who maintain holdings.
Buybacks work very well for executives paid in stock who diversify out of their company stock with suspiciously convenient timing, and speculators.
Sorry, there's no difference between a $100 stock paying a $10 dividend and doing a buyback which increases the stock price to $110. You can always sell $10 to realize the same cash in your pocket.
You know you can hold a stock forever without having to sell it? I have shares that I own from 17 years with absolutely no intention to sell, they pay me in dividends more in one year than I paid for them almost two decades ago.
That doesn't mean you'd be trying to sell for lower prices, does it?
A (sufficiently rational) investor's willingness to sell is mainly a function of current price vs. expected future returns compared to other viable alternatives - which of course includes the dividends paid and any other benefits they might gain from ownership.
And yes, there's also changes in personal need for currency, psychological reasons, etc. but I'd consider these special cases.
Even in your case if you had any good indication towards the stock you're owning being severely overpriced, the likely optimal thing to to would be to sell and dump the money into the next best alternative.
I'd also suggest to be careful in the assumption that "holding a stock forever" is a good idea. Very few companies have existed forever so far.
"A stock buyback is not much different than a dividend payout."
It's very different. In a dividend payout those holding the shares and staying with the company get the money. In a buyback those selling the shares and leaving the company get the money.
When you crunch the numbers buybacks almost always overpay for stock. From an investor's point of view they'd be better with the cash money.
It's another one of those situations where there is a comfortably theory of equivalence that the finance industry loves, that is unfortunately contradicted by the hard facts on the ground. Economics in a nutshell...
Half of the problems of society are caused by this. Once you can attach a number to something, you can make a false equivalence and justify many ills. People often do this unknowingly or even worse, some intend it.
I can expand: They are not actually better at capital allocation than the sum of market participants (obviously, who is?) and so when they lose money it is because the market is so stupid they are losing money, rather than some failure in their ability to model.
“Market can stay stupid longer than you can stay solvent.” that is an argument for not shorting. But you can still abstain from investing, which will be sufficient to outperform those who did invest if you are right.
No one is claiming to be about you time anything which is what’s needed to actually accomplish what you’re asking for. However, if someone says “if you drive at 200 mph at night with your lights off on a winding road in the rain you’ll crash”, you’d be asking “yeah sure, but you can’t tell me which mile marker”.
> “Market can stay stupid longer than you can stay insolvent.” -> Yeah, that is called cope.
You had me at the start but lost me here. That’s not speculation, that’s a fact. Some big shorters of the 2008 crisis almost went broke waiting for the market to catch up.
The adage exists as a reminder that human behaviour can make people run off a metaphorical cliff, and, just like a cartoon character, keep going horizontally until they look down.
But never discount the possibly of spherical cows in a vacuum.
" Right, but what they are supposed to be modeling is the market, so if they went broke doing so they failed at doing that. [0]"
soo now we've moved from "take advantage of investors being stupid by trading against it inidivudally" to "accomplish a FAR HARDER task by modeling the whole market". Cool cool, moving goalposts for days.
Calling certain investors stupid is hardly contradicting the idea that the market as a whole will sort itself out in the end.
Recognizing stupid doesn't mean you are able to do better or shouldn't imply you need to have a better alternative. You can criticize something without being beholden to offer an alternative.
I mean, if you define "being right" as being on the right side of the trade - then by definition it makes sense to "trade against" something if you think you know better.
In reality, it's nowhere near this simplistic. You have to time it. Timing is impossible. Manipulation happens.
No that is something anyone shorting needs to understand fully and deeply.
e.g. the dotcom bubble was clearly a bubble with most dotcoms absolute garbage investments. People looked at it and correctly saw it as such and decided to short. They were 100% correct and still lost big money because it took too long to be proven so.
Dotcoms as they were in 2000 were ridiculous investments that have gone badly south. You couldn't buy google, you could buy Looksmart, repeat that pattern for basically every company that ended up successful. Social media was not a thing. Internet advertising was an utter joke from every level of analysis (is it still? No clue). The thing was clear and obvious and did pan out as expected but man alive it took some time.
In a game of chicken, everyone who plays does so knowing the cliff is right there. Riding what you know is an investment bubble has a little more upside than driving toward a cliff. Same exit strategy, lesser consequence, similarly reckless in attitude.
At the time people showed the necessary cashflows & growth required for profitless doctcom companies to justify their price and nobody could make it make sense. You'd have to be pretty stupid to think greater fools buying can go on forever.
I'm sure some people in finance "believed" in it and others claimed to. Not many in the circles I moved in even paid lip-service unless they were in the IPO business. Plenty played. Did your experience differ?
How do you explain the performance of companies like Lyft, DoorDash, Tesla, etc? People have, rightly, been calling them out as examples of companies that are fundamentally overvalued for years now. But it has only been in the past year or so that the market has actually started to value them appropriately. The market has been extremely irrational since Covid started, but was showing clear signs of that before Covid too.
The title wordplay dates back to at least Drew McDermott's 1976 essay "Artificial Intelligence Meets Natural Stupidity" [0]. The intro is phenomenal.
---
> As a field, artificial intelligence has always been on the border of respectability, and therefore on the border of crackpottery. Many critics <Dreyfus, 1972>, <Lighthill, 1973> have urged that we are over the border. We have been very defensive toward this charge, drawing ourselves up with dignity when it is made and folding the cloak of Science about us. On the other hand, in private, we have been justifiably proud of our willingness to explore weird ideas, because pursuing them is the only way to make progress.
> Unfortunately, the necessity for speculation has combined with the culture of the hacker in computer science <Weizenbaum, 1975> to cripple our self-discipline. In a young field, ,self-discipline is not necessarily a virtue, but we are not getting any younger. In the past few years, our tolerance of sloppy thinking has led us to repeat many mistakes over and over. If we are to retain any credibility, this should stop.
> This paper is an effort to ridicule some of these mistakes.
>In this paper, I have criticized AI researchers very harshly.
Let me express my faith that people in other fields would, on
inspection, be found to suffer from equally bad faults. Most AI
workers are responsible people who are aware of the pitfalls of a
difficult field and produce good work in spite of them. However, to
say anything good about anyone is beyond the scope of this paper.
This paragraph concludes the section titled "Benediction." Ouch.
Between that there is an issue with businesses putting the squeeze on employees because they think AI is ready to replace people even in coding environments. The number of available jobs are shrinking not just because of interest rates, because no one is auditing the natural stupidity of management trying to switch out jobs with AI.
The big firms aren't getting rid of huge numbers of devs, most of the cuts are coming in HR, marketing, and administrative roles.
Start-ups have less funding, so there is going to be less hiring there. We may also see less start-ups in general.
I doubt the boot camps have created an over-supply. There aren't that many boot camps, and from what I understand, the majority of the graduates do not go on to become software engineers. They market based on a few successful grads, but those may be the stand-outs.
The vast majority of "bootcamp types" just add overhead to the time/cost of finding a good hire, I doubt they meaningful contribute to any "market saturation".
I don't think it has anything to do with bootcamp grads. The VC bubble is over and the only way companies can actually survive is to build and deliver solutions that their customers pay actual money for and pay enough of it to turn a profit.
Bragging about "engagement", telling (self-inflicted) war stories about your overcomplicated engineering playground at the next AWS conference, churning through the latest languages and frameworks for no pragmatic reason, or virtue-signalling about "diversity and inclusion" used to yield endless VC money which is how a lot of companies and entire careers started, but nowadays it's back to good old "deliver things someone finds valuable enough to pay for, or die trying".
The end of ZIRP is what is deflating the tech bubble. With interest rates now high(er) than before you can earn a decent return just by buying bonds. There's no reason for investors to recklessly speculate on companies like Lyft or Uber or Doordash when those companies are structurally unprofitable. Throw in the failure of SVB and a couple other banks, and yeah things are going to get very tight in software for the next few years.
At our company, there is a mad dash to build things using LLMs, so that we can turn to the board and say "see? we have LLMs" so that the board can turn to investors and say "see? we have LLMs"
Some of the use cases are certainly of dubious benefit to the actual product.
I'm seeing a lot of the same, and I'd say mostly dubious.
The use cases I've seen that actually make a ton of sense to me are anything with rules engines with cumbersome steps or syntax.
It's so much easier to allow input like 'block traffic from outside the US this weekend from the hours of x to y' or 'junk emails that come from a different smtp server than the domain name indicates' and so on.
Well, you should definitely be able to read the output, agreed on that much.
But sometimes it's annoying to create such rules either due to having to click through multiple screens, or not remembering the exact syntax offhand.
Locally, something like iptables comes to mind. I've used it hundreds of times, but never regularly enough to remember how to write more complex rules without consulting the manual or Google each time.
The way I'd do it would be to have the LLM generate a rule in some fairly-readable rules engine language, and show the rule, examples, and counter-examples for the user to confirm.
This would be like writing email filters in gmail or smart playlists in iTunes.
Possibly, you could ask the LLM for the examples and counter-examples that look most likely to be a bug?
I think realistically, you can't use AI for anything where the output absolutely must be precise. AI is great for when close enough is good enough, but even just specifying "close enough" is hard.
If you need to know that the output is perfect then you need to check it, and if you need to check it then maybe it's better to just get an expert to do it in the first place.
Again, I'm talking about in situations where you need to be absolutely precise. This isn't for emails explaining simple concepts to a client (which is a good use) but rather a technical document explaining a process to another expert end user.
Essentially, yes. A lot of SaaS companies provide their own syntax, workflows, or PL, that you as an end user don't use enough to be perhaps know the exact syntax offhand. Even something like AWS and their policy documents, for example. And don't forget something like AWS/Terraform.
I'm no AI hypester. But I have made some observations over my "N" decades on this planet so far. That a lot of humans are either:
1. dumb
2. ignorant
3. crazy
4. hateful
5. violent
6. biased
7. indoctrinated
8. flaky
9. dishonest
10. fabulists
11. bored
12. lazy
13. conspiring
And usually an ever-churning mix of many of them, all at once.
Therefore, to make a human-equiv AI its not necessarily "incorrect" to give them those traits too. NOT saying that would be useful, or a desirable ideal to strive for. But it WOULD make an AI act more like an authentic human!
I walk the talk too.
How so? For years now I've been writing a little story serial (DSPR) featuring an AI character who has all those qualities. Sometimes all evidenced in the same single conversation. And lately have been building a game featuring the same kind of insights, and shenanigans. (Slartboz -- see my bio if curious for more details.)
An absolutely huge part of it for humans is that as the complexity of the thing being reasoned about grows, the accuracy of the mental models declines. That and we also train billions of individual models on a narrow slice of reality, and leverage a distributed consensus protocol through low-bandwidth "conversations" to update the weights. I think you could probably emulate this with conversational agents each with their own LoRA, that way they wind up with differing opinions.
The trick isn’t getting out when you realize the grift. The trick is getting out shortly before everyone else does.
Actually, the trick is getting out before everyone getting out before everyone else does before anyone else does. (And so on for higher values of recursion.)
Look at the real estate bubble that happened in China. They were literally building uninhabitable houses, but people invested in them with the knowledge that someone else would still buy them.
It doesn’t take a genius to realize uninhabitable houses don’t have real underlying value.
Reuters was the first news agency to use NLP to automate news generation back in the early 2000s. They have always had a pretty robust NLP/ML group and the author of the article is an ex-Fund Manager who is writing an op-ed aimed at other Fund Managers/PEs/less technically knowledagble investors.
Tiger Global and SoftBank made similar mistakes in the Deep Learning revolution because they were Fund Manager types with a lack of domain experience, compared to say A16Z, Sequoia, Greylock, etc.
An Investment Associate at BVP or an MLE or ML PM looking at making their own Generative AI startup with YC isn't the target audience - institutional investors who lack techncial domain experience are the target audience of this article
Hat tip for this post. I assumed it was a regular news article from Reuters. It is not. I would classify it as an editorial by a professional money manager. Here is the author: https://www.breakingviews.com/journalist/felix-martin/
I help run a microbiology / phage biology conference. Our team is tiny and volunteer. Previous years we needed an army of interns / undergrads to do a lot of the work; this year we're using an Army of Interns. Tasks include:
- doing a baseline check for submitted abstracts
- summarizing / listing out key points of abstracts
- creating visa letter drafts for international travelers
- creating invoice documents for large groups
- abstract submissions are done via an optional upload > extract > form filling workflow with an LLM, where it just fills in the form for you (things like author names, correspondence, abstract text, etc.) which saves a bunch of time
Basically I have code (sometimes AI, sometimes not) doing the heavy lifting here, instead of bossing around a bunch of undergrads. Undergrads are also prone to getting distracted, hallucinating facts, making mistakes, etc. so it's not so different than before, except we save gobs and gobs of time.
Oh man, that reminds me that I used to have to read engineering patents as an intern. I think AI would be much better at what I was trying to do, or at least provide a better starting point than the actual patents.
The usual midwit gold rush is annoying, but, in practice, beneficial to all of us.
This kind of Cambrian explosion of copycats and wannabe is simply accelerating the hype cycle, we won’t have to wait a decade to find out if this tech is actually good.
The only good news is that interest rates are likely to stay high for the next few years, which I expect will put a major damper on this bubble. AI companies will have to show profitability quick in order to keep the funding train going.
I don't understand how you can in 2023 have any economic theory that has "market actors act rationally in their best interest" as an underlying assumption.
196 comments
[ 2.9 ms ] story [ 254 ms ] threadGenerative AI is very useful for creating rigged demos, which are particularly useful when people want to believe. Widespread skepticism might have some effect, but there are always skeptics, and no particular reason to believe there won’t be any winners to make venture capitalists happy with a few of their bets, enough to pay for the failures.
The only access investors have to outsized gains are through investing in the companies that invest in these companies. What small investors need are private equity index funds.
Someone else mentions Netflix vs. Blockbuster. Small investors would have been doubly screwed if they could only invest in Blockbuster because Netflix was still private.
The number of (legitimate) publicly-traded internet-related companies in the late 90's that were accessible to the average investor was also incredibly small. Legitimate small companies are rarely traded on exchanges in general, as they have no reason to solicit investment from the general public. It's the same phenomenon behind crowdfunding; if you have a legitimate product in development, you can simply convince a conventional lender to loan you the initial capital.
The fact that investors are constantly chasing new and risky ideas, instead of just zero sum assets, is why the US has the best jobs market in the world.
If investors and consumers were responsible and conservative with their money, dev wages would be much much lower. So enjoy it, because you have it good.
Of course, none of them ever do.
Investing your own money is a heluva lot different, and it's where the rubber meets the road.
I'd prefer multiple fashions in play at once.
Nobody is forcing you to work at FAANGs or other hypergrowth companies that have redundancies, there's older established companies that didn't have redundancies which of course, pay much less than FAANGs do.
Places that only have safe employment options, without hypergrowth, also have massively lower wages for everyone.
When market contractions hit even "safe jobs" aren't necessarily safe.
Competition is important. But there are limits, after which it's better for everyone if the business-creator comes up with a novel idea and pursues that instead of iterating on everyone else. This is very difficult. It might be a bit less difficult if the culture didn't encourage followership in its leaders and funders.
The Duke of Westminster for example was the youngest billionaire in the UK with an estimated net worth of £10bn at 35. His family's land includes 100s of acres of the most valuable real estate in London. Their ancestor was granted these lands by William the Conqueror in 1066.
Of course, neither are European investors, but that's not the point. The point is a dinamic jobs market where the money goes on ideas and businesses, insted of assets, pushes all wages up and benefits the middle class.
Investors living off land and other such rentseeking endeavors destroys the middle class as it's just feudalism with extra steps.
>There is little benefit to be gained from disposing of that wealth to speculate on the latest fad and facing taxes.
Of course, but this is the fault of our tax system which incentivizes land hoarding and rent seeking by those already afluent, at the expense of the have-nots looking to buy in.
I let out a family house. The rent is the greater part of my pension. I used to live there, until I was divorced (it was too big for one person). It's been clear to me for decades that freehold land was a better investment (here) than investing in a conventional pension; and I lived through the outright theft by Robert Maxwell of the whole of his staff's occupational pension fund. So I paid off my mortgage as quickly as I could.
Regarding the chores, that someone mentioned upthread: I have a long-term tenant, a family. Their kids go to school around the corner. They mow the lawn. They always pay the rent by bank transfer. I hire tradesmen to deal with the gas, plumbing and electricals. I visit twice a year to check that everything's OK. These aren't really problems.
And rental housing is scarce where I live. It's a university town with several important hospitals, and surrounded by a no-development green belt. I'm providing a social benefit.
I'm not sure when this kind of "rent-seeking" turns into the kind of rent-seeking of which people disapprove.
Where I live there are individuals owning entire apartment buildings, >10 houses, in fact, 80% of the entire country's wealth is owned by the top 300 richest individuals.
Single individuals renting out their spare home, doesn't even register on the radar compared to what the rentseeking of the truly wealthy looks like.
An example of rent seeking would be if you lobbied the local council to restrict new building or additional rental units in the area leading to artificially higher rents. It has nothing, directly, to do with receiving income for renting a housing unit
There are many sides to this. Over a decade ago, Apple colluded with other tech companies to artificially suppress wages by crafting "no-poaching" agreements. Facebook refused to play ball.
Nope.
There's:
1. property tax
2. insurance
3. maintenance
4. upgrades
5. dealing with the lawn
6. 6% commission taken by the real estate agent
7. inflation
8. all the time spent dealing with it
9. mortgage interest
10. the costs from having it sit empty on the market waiting for a buyer
11. and the most neglected cost, the time value of money
Properties here are sacred and the tax incentives, tax breaks, basically zero inheritance taxes, make it one of the safest kind of investments.
And that's without the scummy practices of the central banks and governments who manipulate the market to keep prices going up.
If you just let the property rot without a tenant, then, yeah, this punishes you. I'd recommend against letting a property rot without a tenant, just like I'd recommend against buying 500 barrels of crude oil, and letting them rot away into worthless sludge in your cellar over twenty years. All things that we buy have a limited shelf life, you need to use and maintain them before they expire. Structures built on a property are no different.
#6-11 are real costs of investing into property, and I agree - they cost a hell of a lot more time and money than buying an index fund and sleeping on it for 40 years.
But the real problem with real estate "investment" is that the return doesn't come from doing something productive but rather is predicated on preventing someone else from doing something productive, namely making more buildings.
The really big winners were those which were undervalued because nobody expected them to grow as much as it would. Back in the 90s: "A mail order rental service growing bigger than Blockbuster? Get real." Being rational would filter the successful outliers along with the vast field of bad ideas. Usually going unfiltered is a losing proposition like playing the lottery. However, if the wins are outsized enough it perversely "pays to be stupid" assuming you have enough capital to win once and of course that similar market conditions emerge again.
That said useful applications of AI which may be implemented relatively easily are clear big winners as investments but that is pretty tautological and unhelpful for actually figuring out if say, an "AI love letter writing service" would be a terrible idea or not.
For every company that hit the niche and made billions over it there is few that were too early, and few that were not lucky enough to get the opportunity that made the winner win.
OpenAI is selling pickaxes. It's also selling fear in every color. Fear for your future. Fear for humanity's future. Fear for the future. It's selling hope - of new profits, new discoveries, new futures. It's selling prophesy and it's selling religion.
It's also selling entertainment. To be honest, I bet that majority of use of ChatGPT still classifies more as fun than work. I pay for GPT-4 twice (ChatGPT+ and pay-as-you-go API access), and I use it daily, and 90% of my use is... exploratory in nature. I still consider it money well-spent. The problems may be real, and the outputs may be useful, but the choice of using GPT-4 is still dictated mostly by... playful curiosity.
(Also, it's the first tech thing in almost a decade that broke through my layers of cynicism and got me excited again. There's lot of cool things to play with, lot of discoveries to be made, and the progress is real.)
Whether or not LLMs are intelligent, OpenAI won. Financially, for now, but they'll have a spot in world history (for as long as history is a thing). The genuine technological jump they introduced... I can't remember any tech company in my life making this big a change in one step.
How old are you?
May 5th, 2023: http://www.threepanelsoul.com/comic/new-paradigms
An investing payout sometimes creates something of value for society: when the thing invested in turns out to be useful to people.
Gambling: the expected ROI is negative
Investing: the expected ROI is positive
Even without any gambling I have a quite back track record of getting back the money I gave away as loans.
Some days it's wisest to wait until you're ready.
In a sibling post I talked about Google's inability to spend, this is also true for old school Blue chip companies like say Oil& Gas, they have had enormous record profits last couple of years and while they have for good reasons not doing lot of new wells[1] they have not not also done much green energy investments either.
[1] Financially sound not do so, given the long term prospects of oil are not great, and short term price fluctuations are not good decision basis for 30+ year projects new drilling entails.
The only practical difference is you pay taxes on the dividends when you receive them, and taxes on the capital gains when you sell.
Investors expecting a return on their investments? How awful!
There's a bill in Congress to "add more accountability" to stock buybacks by imposing taxes (https://www.congress.gov/bill/117th-congress/senate-bill/275...), but it's clear congress has no interest in curtailing this practice unfortunately.
And why would anyone invest in a company that will never pay dividends/do buy-backs?
The result of buybacks is that the number of shares drops. E.g. Berkshire's stake in Apple raised from 5.2 to 5.5 % merely as a result of Apple's buybacks.
Obviously shareholders benefit from having a larger piece of the cake while not requiring to pay any tax on this operations.
It depends entirely on whether it is an ordinary or a qualified dividend. This is true in the US, other jurisdictions vary
It's simply not a tax dodge.
If I bought a stock yesterday and they issued a dividend today, there is no way for me to avoid the higher tax rate.
If I bought it yesterday and they decide to do a buyback today, I can hold it until the lower capital gains rate kicks in, thereby avoiding the higher tax rate.
See how I can avoid paying the higher tax rate?
Play some more games with tax loss harvesting, and the buyback option gives you a lot more ability to pay lower taxes. With the dividend there is a limit to the amount of loss you can offSet with short term losses.
Simply put, in any given situation capital gains is likely to give you a better option to pay lower taxes.
In other words, buybacks are a way of paying way less tax.
Not way less, just choosing _when_ to pay a tax (say, during retirement years instead of working years of one's life).
I don't see anything wrong with doing this.
Buybacks work very well for executives paid in stock who diversify out of their company stock with suspiciously convenient timing, and speculators.
> and speculators
Everybody who invests is a "speculator".
Investors on the other hand are not necessarily interested in price fluctuation but owning their share of the cash flow through growth or dividends.
A (sufficiently rational) investor's willingness to sell is mainly a function of current price vs. expected future returns compared to other viable alternatives - which of course includes the dividends paid and any other benefits they might gain from ownership. And yes, there's also changes in personal need for currency, psychological reasons, etc. but I'd consider these special cases.
Even in your case if you had any good indication towards the stock you're owning being severely overpriced, the likely optimal thing to to would be to sell and dump the money into the next best alternative.
I'd also suggest to be careful in the assumption that "holding a stock forever" is a good idea. Very few companies have existed forever so far.
I still haven't seen the actual argument for what would make a difference when it comes to Dividends vs. Buybacks.
to some people, it's a psychological difference, not a financial difference.
Could you explain why you said "nope"?
It's very different. In a dividend payout those holding the shares and staying with the company get the money. In a buyback those selling the shares and leaving the company get the money.
When you crunch the numbers buybacks almost always overpay for stock. From an investor's point of view they'd be better with the cash money.
It's another one of those situations where there is a comfortably theory of equivalence that the finance industry loves, that is unfortunately contradicted by the hard facts on the ground. Economics in a nutshell...
It's a stubborn belief in 'efficient markets'.
“Market can stay stupid longer than you can stay solvent.” -> Yeah, that is called cope.
You had me at the start but lost me here. That’s not speculation, that’s a fact. Some big shorters of the 2008 crisis almost went broke waiting for the market to catch up.
> Some big shorters of the 2008 crisis almost went broke waiting for the market to catch up.
At what point are you a successful predictor of the 2008 crisis vs. a guy who just shorted the economy in 2006?
[0]: https://www.econlib.org/archives/2015/08/recessions_ofte.htm...
But never discount the possibly of spherical cows in a vacuum.
soo now we've moved from "take advantage of investors being stupid by trading against it inidivudally" to "accomplish a FAR HARDER task by modeling the whole market". Cool cool, moving goalposts for days.
Calling certain investors stupid is hardly contradicting the idea that the market as a whole will sort itself out in the end.
Like on a linked list, you can only "point" to the next big disaster!
In reality, it's nowhere near this simplistic. You have to time it. Timing is impossible. Manipulation happens.
No that is something anyone shorting needs to understand fully and deeply.
e.g. the dotcom bubble was clearly a bubble with most dotcoms absolute garbage investments. People looked at it and correctly saw it as such and decided to short. They were 100% correct and still lost big money because it took too long to be proven so.
Dotcoms as they were in 2000 were ridiculous investments that have gone badly south. You couldn't buy google, you could buy Looksmart, repeat that pattern for basically every company that ended up successful. Social media was not a thing. Internet advertising was an utter joke from every level of analysis (is it still? No clue). The thing was clear and obvious and did pan out as expected but man alive it took some time.
If everyone knew it was clearly a bubble, it would not be a bubble.
At the time people showed the necessary cashflows & growth required for profitless doctcom companies to justify their price and nobody could make it make sense. You'd have to be pretty stupid to think greater fools buying can go on forever.
I'm sure some people in finance "believed" in it and others claimed to. Not many in the circles I moved in even paid lip-service unless they were in the IPO business. Plenty played. Did your experience differ?
---
> As a field, artificial intelligence has always been on the border of respectability, and therefore on the border of crackpottery. Many critics <Dreyfus, 1972>, <Lighthill, 1973> have urged that we are over the border. We have been very defensive toward this charge, drawing ourselves up with dignity when it is made and folding the cloak of Science about us. On the other hand, in private, we have been justifiably proud of our willingness to explore weird ideas, because pursuing them is the only way to make progress.
> Unfortunately, the necessity for speculation has combined with the culture of the hacker in computer science <Weizenbaum, 1975> to cripple our self-discipline. In a young field, ,self-discipline is not necessarily a virtue, but we are not getting any younger. In the past few years, our tolerance of sloppy thinking has led us to repeat many mistakes over and over. If we are to retain any credibility, this should stop.
> This paper is an effort to ridicule some of these mistakes.
---
[0]: Drew McDermott. 1976. Artificial intelligence meets natural stupidity. SIGART Bull., 57 (April 1976), 4–9. https://doi.org/10.1145/1045339.1045340
This paragraph concludes the section titled "Benediction." Ouch.
> Natural stupidity beats artificial intelligence every time.
I think it’s more likely the boot camp type that are over saturating the market. Just supply and demand.
The big firms aren't getting rid of huge numbers of devs, most of the cuts are coming in HR, marketing, and administrative roles.
Start-ups have less funding, so there is going to be less hiring there. We may also see less start-ups in general.
I doubt the boot camps have created an over-supply. There aren't that many boot camps, and from what I understand, the majority of the graduates do not go on to become software engineers. They market based on a few successful grads, but those may be the stand-outs.
Bragging about "engagement", telling (self-inflicted) war stories about your overcomplicated engineering playground at the next AWS conference, churning through the latest languages and frameworks for no pragmatic reason, or virtue-signalling about "diversity and inclusion" used to yield endless VC money which is how a lot of companies and entire careers started, but nowadays it's back to good old "deliver things someone finds valuable enough to pay for, or die trying".
I don't think the software industry is in a recession.
Some of the use cases are certainly of dubious benefit to the actual product.
The use cases I've seen that actually make a ton of sense to me are anything with rules engines with cumbersome steps or syntax.
It's so much easier to allow input like 'block traffic from outside the US this weekend from the hours of x to y' or 'junk emails that come from a different smtp server than the domain name indicates' and so on.
But sometimes it's annoying to create such rules either due to having to click through multiple screens, or not remembering the exact syntax offhand.
Locally, something like iptables comes to mind. I've used it hundreds of times, but never regularly enough to remember how to write more complex rules without consulting the manual or Google each time.
This would be like writing email filters in gmail or smart playlists in iTunes.
Possibly, you could ask the LLM for the examples and counter-examples that look most likely to be a bug?
If you need to know that the output is perfect then you need to check it, and if you need to check it then maybe it's better to just get an expert to do it in the first place.
Yeah but also maybe an expert gets a shipload more done using an AI? Huge productivity increase? Dunno!
As long as the mad dash translates into healthy 8-hour work days - have fun!
Investors get a nice presentation. Executives get to be part of the trend for marketing. Employees learn about the latest technology
Note no one is proposing replacing management with AI, just the rank and file.
So far what we’ve been able to build feels brittle. But, LLMs are fun to play with though
1. dumb
2. ignorant
3. crazy
4. hateful
5. violent
6. biased
7. indoctrinated
8. flaky
9. dishonest
10. fabulists
11. bored
12. lazy
13. conspiring
And usually an ever-churning mix of many of them, all at once.
Therefore, to make a human-equiv AI its not necessarily "incorrect" to give them those traits too. NOT saying that would be useful, or a desirable ideal to strive for. But it WOULD make an AI act more like an authentic human!
I walk the talk too.
How so? For years now I've been writing a little story serial (DSPR) featuring an AI character who has all those qualities. Sometimes all evidenced in the same single conversation. And lately have been building a game featuring the same kind of insights, and shenanigans. (Slartboz -- see my bio if curious for more details.)
Its like the system truly needs to meltdown and actual VCs and POEs need to lose their shirts to get a hint.
Actually, the trick is getting out before everyone getting out before everyone else does before anyone else does. (And so on for higher values of recursion.)
Look at the real estate bubble that happened in China. They were literally building uninhabitable houses, but people invested in them with the knowledge that someone else would still buy them.
It doesn’t take a genius to realize uninhabitable houses don’t have real underlying value.
If there is societal pressure to not invest in something it is, by definition, undervalued.
Reuters was the first news agency to use NLP to automate news generation back in the early 2000s. They have always had a pretty robust NLP/ML group and the author of the article is an ex-Fund Manager who is writing an op-ed aimed at other Fund Managers/PEs/less technically knowledagble investors.
Tiger Global and SoftBank made similar mistakes in the Deep Learning revolution because they were Fund Manager types with a lack of domain experience, compared to say A16Z, Sequoia, Greylock, etc.
An Investment Associate at BVP or an MLE or ML PM looking at making their own Generative AI startup with YC isn't the target audience - institutional investors who lack techncial domain experience are the target audience of this article
And: https://www.1167capital.com/bio-felix-martin/
He sounds like the real deal.
- doing a baseline check for submitted abstracts - summarizing / listing out key points of abstracts - creating visa letter drafts for international travelers - creating invoice documents for large groups - abstract submissions are done via an optional upload > extract > form filling workflow with an LLM, where it just fills in the form for you (things like author names, correspondence, abstract text, etc.) which saves a bunch of time
Basically I have code (sometimes AI, sometimes not) doing the heavy lifting here, instead of bossing around a bunch of undergrads. Undergrads are also prone to getting distracted, hallucinating facts, making mistakes, etc. so it's not so different than before, except we save gobs and gobs of time.
This kind of Cambrian explosion of copycats and wannabe is simply accelerating the hype cycle, we won’t have to wait a decade to find out if this tech is actually good.
[1] https://en.wikipedia.org/wiki/Gartner_hype_cycle
however, chatgpt and heypi have answered the question of learning syntax at scale. i am impressed.