Alpha is ultimately the result of analysis, of better analysis than others.
LLM's can actually be exceptionally good at research and pattern recognition, i.e. analysis. And while they aren't great at running numbers themselves, they can do exceptional work passing off Python scripts to an interpreter to generate the numerical results they need.
I'm quite sure the Robinhood AI is going to be trash, i.e. just a gimmick.
But, it's not crazy to think that with the right harness, there are big opportunities for identifying profitable strategies. Especially relying on unparalleled and essentially unlimited research capacity based on public information. More analysis than any single firm could ever hire.
And even for Robinhood users, it's entirely plausible that AI-traded stocks will perform much better than the trades a majority of users would make, since most investors are really unsophisticated.
>LLM's can actually be exceptionally good at research and pattern recognition, i.e. analysis.
No they aren't, they're good at imitating analysis based on representations of analysis in their training data. Also, Its likely that out dated techniques would over represented in training data.
Do you think Jane Street would have the returns they do if they just imitated all their competitors and everyone was using the same strategies?
That sounds like more training data that the human is just regurgitating. Nobody I know has ever had an original thought, just combined existing thoughts that were in their training data, in new combinations.
Thats a sad way to perceive the human experience and the individual. Its lowkey a techno fascist psyop you're falling for if you believe this to be true.
Well said. Lots miss this key point and it’s what I mean by alpha. Where to start, what the angle is, and LLM can help develop though I would of course argue it’s a fools errand with how crowded the space is but you could still develop an idea but it’s not going to create the idea for you.
Yeah if you have an actual really good idea for a certain kind of under analyzed data that could give you an edge but it would take too much effort or time to compile or analyze the data, an LLM has potential to make a viable strategy out of an otherwise correct but unuseful insight
I mean sure an LLM can indeed have useful ML functions BUT it is a fools errand to think an LLM at the retail level is generating any alpha. Help build a model sure. Whatever these weird mechanism of agents trading, no.
If someone is poor and doesn't have the money, sure you can't enforce it.
If it's a corporation, it's pretty straightforward. If they refuse to pay, you get a writ from the court that authorizes seizure of assets.
Usually that means you go their bank and the value of the judgment will be garnished by their bank and given to you.
Occasionally and theatrically, a sheriff will take you to their headquarters to seize property like computers and printers that you can sell at auction until the value is satisfied.
It only becomes difficult if the corporation is bankrupt, which is similar to a poor person who doesn't have the money. Then it becomes a question of prioritization, e.g. do you get paid before or after lenders, and will there be any money left.
Have you actually gone through this process? Like sure, obtaining a writ is technically part of the same case, but it's pretty much starting all over again. And you'll still be paying filing fees, dealing with court clerks, and waiting weeks or months.
Finding a corporation's bank is a whole separate issue, where you have to go back to court for a post-judgement discovery to force them to tell you. And even if they do - or you already knew - you have to get the writ served to the bank, and just hope they didn't move funds beforehand - or else you're back to start.
As GP said, it IS a huge PITA to get judgments paid, and it's particularly menacing in Small Claims. Unless the other side act on some virtue (which, they were already bad-faithed enough to have a lawsuit against them AND lose), your judgment is just an IOU, and actually forcing collection is often way more money — or time in money — than most state's Small Claims limits.
No, but my point is that it does work. It's a lot better to be delayed by months and have to pay more but still get your money, rather than by delayed by years and not get it at all.
This is a franchised retailer with over 300 locations, and this is a value of $200,000+ plus so this is way bigger than small claims.
Like I'd definitely agree with you if we were talking about a $5K claim against a single location in small claims. But this seems to be a $200,000+ claim against a corporation in regular court, as far as I can tell.
I mean yeah, but only my final paragraph is about small claims. The accountability and enforcement mechanisms to collect your money from a ruling don't change between a $200K and a $2K lawsuit, it's the same PITA either way - just a few more zeroes on paper.
The store in question was a franchise so potentially the liability will be limited to just the assets of that franchise. But there's a lot of weird stuff here and it looks like the corporation may have (in a legally questionable manner) removed assets from the store to the corporation during proceedings to shield them.
A representative of the corporation, while taking over the store, expressly states that the corporation is taking over the consignment agreement, on camera and with several witnesses.
They absolutely are and a good lawyer, I'm sure, could audit the accounts and find some misdeeds - the issue is that the auditing and even getting access to those records in court is extremely expensive. To my knowledge there isn't a way to trigger that kind of a discovery in small claims so you need to go through the pricey legal system.
The money in question here is the proceeds from selling a collection valued at 200k - the recovery (unless you start to get into punitive territory) is likely to be rather meager... and it's a large risk so there may be few bites on firms willing to take it on purely commission.
In the UK you can make an order of information to compel directors of a company that is in debt to answer questions about company assets, accounts and records under oath. This can be done in County Court and my understanding it is inexpensive. I'm not sure how useful this is for carrying out an audit because I think its meant to be used for seeing if the debtor has the ability to pay. I think generally incorrect trading during an insolvency is meant to be discovered by the receivers during the insolvency process. Also, I'm not sure if there is an equivalent to an order of information in the US system.
the issue is that the auditing and even getting access to those records in court is extremely expensive.
In most cases, the bankruptcy trustee will be doing that work already.
But in a case like this, it's probably not going to be necessary. Courts usually pierce the corporate veil in situations involving the debts of wholly-owned subsidiaries. It happens frequently enough that its actually news when they don't pierce the veil. This is because corporations usually do a bad job of doing all the things that are necessary to maintain the liability shield in court.
In a nutshell: it requires treating the subsidiary as an entirely separate entity, with separate books, accounts, back office, officers/management, etc. As this is extremely inefficient, most corporations don't bother. The only corporations that do are the ones that deal with company-killing litigation regularly enough that it's worth it to absorb the cost of maintaining the liability shield.
Somewhere in one of the long videos, they mentioned that there were unique stickers on each of his items that he was selling on consignment. They had to have known which items were his.
The original franchisee claims to have lost their life savings in that move. I have no idea how exactly that happened. Their story really sounds like something from Russia back when western investors had their company simply taken from them by someone well-connected.
It sounds like the original franchisee doesn’t want to admit that they were losing a lot of money already. Only someone really desperate would take on a $200K lego collection and only collect a 10% consignment fee. It would also explain the corporate “takeover” if they were already behind on paying their franchise fees or whatever they might have owed to corporate.
That being said, it’s not illegal to be a bad business person, and none of that excuses the subsequent behavior by BAM corporate or the new franchise owner.
* They had sold about half of the consigned inventory
* The old franchise owner said they had a job offer outside the country
* Said franchise owner was current on a restructured fee schedule that, they allege, was the direct result of corporate bungling the transfer of accounts from the franchise owner previous to them
I definitely heard 10% first and only later 35%. For some reason the videos don't have transcripts and the Gemini AI isn't available, so I can't try to search for it. But I'm 100% sure that 10% was the figure mentioned first (maybe I misunderstood and it was just being used as an example). If the real figure is 35%, then I retract any comments about them making a bad business deal.
From what I understand, the original franchisee wanted to sell the store because they wanted to leave the US (for "political reasons"; I suspect they don't want to live in Trumpland anymore, but that's pure speculation). The way it appears, the moment they announced that desire to sell, B&M corporate showed up to take control of the shop. And the consignment.
> Then Bricks & Minifigs Corporate took control of the Salem location from the original franchise owner
> They were found liable in court. They closed the store rather than pay.
This doesn't make any sense. If the corporation took control of the franchise, the corporation now owns it and its obligations. They can close the store if they want, but that doesn't do anything about their obligation to pay.
What's missing from this story? Because as presented, it makes no sense.
It is my understanding that they sued the franchise, which then closed, and not the parent corporate entity.
While they won against the franchise due to the default judgement, they didn't win against corporate.
The store that is now closed is the franchise they sued.
Right. I was very surprised because there aren't laws against insider trading on prediction markets.
So I genuinely don't understand what he is being charged with. What precisely is the "fraud"? The entire point of prediction markets is to get people with better information to participate, i.e. "insiders".
Insider trading with public corporations has tons of specific laws around it to clearly define what is insider information and what isn't. Prediction markets don't have any of that.
And the article does nothing whatsoever to clarify what the heck the actual fraud is supposed to be.
(And I understand this is against Google policy, but that's not what this is about.)
> but it's not Google Search, and when I want Search I want Search.
Not me. I really appreciate having both results simultaneously. I can scan the first couple sentences of the AI response, and if that already has the answer then great. I can expand it to see if there's more.
Or, if I see that the AI mode didn't understand my brief search query, I just glance at the search results below.
And often times, when I do need to follow a link, I find the source result links in the AI mode to be a better quality than the search result links.
Most of the time I'm looking for something very specific that there are plenty of articles about, but clicking on the articles results in popups, banners and an unhealthy amount of scrolling to get to the answer.
AI overview provides me the answer instantly.
Think about suff like "does china borders afghanistan".
In those cases you can be confident that the AI overview is right, and saved you time.
If it is a complex or niche question I tend not to trust the overview and go straight for legitimate-looking results
The LLM results are presented confidently and succinctly in a way that is designed to tell you “yes” OR, it not applicable, it just mashes together statements (which often leads to a response that contradicts itself one sentence later). That’s not the same as your vetting search results.
Well before Google screwed it all up there used to be some correlation between top hits and what you were looking for. SEO has muddied the waters for many years now and it’s never been truly “merit based” or “objective” or whatever we want to call it, but generally speaking, the first results were the best by default.
Ok, but it’s been mine. And clearly I’m not alone.
I feel like at this point any discussion about LLM’s has an implied “my experience” because LLM’s are super inconsistent due to not being refined tools at all. I’m sure your experience has been different, just like my experience has been different. I imagine you’ll want to chalk it up to operator error, but it sure seems like a lot of people have variations of my experience. If so many people are operating it wrong, then maybe the tool is poorly designed.
Understand that I use LLM’s pretty frequently. I am not “anti-AI.” I’ve used production tools incorporating machine learning for years now. But LLM’s simply aren’t the bespoke tools that these companies want you to believe, and they are definitely not a suitable replacement for search. It’s simply too inconsistent and will hallucinate answers. Google search didn't make up answers, it presented indexed sources that you ver in real time which I find to be a far superior way to do research. I don’t like having to guess when an LLM is just making shit up as it asserts something with simulated extreme confidence. Not only that, you can take a correct answer from a LLM and just start saying “know that is not right,” and it will start apologizing to you and generating other answers. That is a huge problem! I shouldn’t be able to “convince it” to give me a different answer.
Yes SEO made things objectively worse. Doesn’t mean we need to add another layer of issues on top of that.
It replaced some of my most used tools with google search. I used to be able to search "define inoculant" and I would get a definition, synonyms, and even a history of the word usage. Now it's replaced by an often mistaken AI summary. Even "inoculant synonyms" doesn't work.
This and the the other thread that talks about RL and synthetic data seem to suggest that AI can figure out all the technical issues without humans looking into them. I'm not sure if that's true at all.
That assumes there is documentation or examples. A big reason Stack Overflow took off was people struggling with things like the Android API documentation.
Some of those discussions made people go figure out how to do it, and then post it as an answer. The knowledge didn't exist anywhere until they did.
It might make sense for AI companies to throw agents at new technologies to trial-and-error their way to internal documentation which they then provide to their models. On the other hand, the people making tomorrow's APIs have LLMs too and that makes documentation ~free. Hallucinations could still bring you back to the first hand, though.
Sounds nothing like the world we live in. When has there ever been a time where there were an abundance of software documentation? How can plenty of documentation or code be made if AI scraper bots hammer servers that host them, steal content and drive people away from the actual authors?
I think it's fascinating because it seems to be a completely different type of compression.
You can see it in the hair as well. It seems very clear that it is engaging in a kind of texture synthesis.
So it seems to be looking at an area, and capturing the textural quality. And then reproducing that, so the overall effect is the same, but individual fibers or fuzzy bits are randomly generated from scratch.
And so yes, if you zoom in enough, the knitting looks completely wrong because the regular geometric pattern of irregular yarn it is made of has been replaced by a completely irregular pattern of irregular yarn.
In other words, it is essentially hallucination of details on a micro scale but not on a macro scale.
And I think that raises a really interesting philosophical question of what we consider to be valid image reconstruction from lossy compression.
Because on the one hand, this is no different from blurriness or even the kind of blocky JPEG compression we are familiar with. It's just pixels that are wrong. Those blocks don't appear in the original image. The blurriness isn't there in the original image.
But on the other hand, we see blurriness as being somehow more "honest", and we are easily able to recognize that blockiness is an artifact.
Whereas with textural hallucination, it is no longer clear what is being filled in versus what is original, because it's doing such a good job of emulating so many aspects of the original texture.
And it's really hard to say if one approach is better or worse than the other. It's probably more accurate to say that one is more appropriate than the other in different contexts. Like if it is just a normal news photograph, I am perfectly happy with a sharper image because it's not changing anything substantial – it's not changing the face of a world leader or the number of people in the photo. But on the other hand, if I am doing online shopping for shirts and I want to be able to zoom in on the texture, then it's incredibly important that the texture be accurate and not loosely hallucinated.
This is a potential problem in "AI" denoising as well.
These denoising models, the autoencoders more directly so, work by (lossily) mapping the raw input to a very low dimensional representation. The other part generates the desired image back from the low-d representation.
The problem is that nothing, in the vanilla versions, prevent the the low-d version to be a semantics representation such as, Moon, dark hair etc and the generative part to take cues from the semantic representation to a generated sub-image.
The Samsung phone Moon image was likely a result of deliberate choice / company policy, but these things can happen without explicit intent.
I disagree that it's only on a micro scale. If you look at the picture of the parrots it completely changes the black/white pattern in the face of the red parrot and if you look at the picture of the green bicycle where the luggage rack attaches close to the center of the rear wheel, it's completely mangled, in contrast to the more "blurry" picture where you can clearly see the bolts where it's attached also the rods going from the wheel hub up to the luggage rack also looks very jagged and weird whereas they look fine in the blurry one. There are certainly other errors as well but those where the most jarring I Noticed at a quick glance. I don't think a compression algorithm that does this poorly on cherry picked examples are going to fly when you start throwing real pictures at them. If you are going to screw with the ground truth I bet you could get better results by throwing the blurry pictures in one of those "AI" upscalers.
I would say all of those examples you are picking are at the micro scale. Obviously it's a somewhat arbitrary division between macro and micro, what you consider to be the macro objects versus what you consider to be the micro details.
And this is also going to depend on the level of compression being chosen. Obviously, the greater the compression, the lesser the fidelity. The lesser the compression, the greater the fidelity.
Camera phones have been hallucinating texture and details for years now. This is nothing new, it's just now part of the compression layer as well.
And defense attorneys have been making arguments about the unreliability of all sorts of types of evidence for many centuries. So there is nothing new there either.
If someone's face is clearly visible and recognizable in a photo, this algorithm isn't changing their face to someone else's face.
That's just a snarky way to describe all business investment that requires purchasing things made in the future. It's entirely normal.
You could literally rewrite the quote to be about iron and about building railroads for trains and passengers that don't exist yet. See how silly that would be?
Except the "profit that is mathematically impossible" part. That's just made up and false. It's entirely possible that we are actually underestimating demand, and there is going to be tons of profit. Nobody knows for sure, but profit is very, very, very possible.
> You could literally rewrite the quote to be about iron and about building railroads for trains and passengers that don't exist yet. See how silly that would be?
My point is, railroads turned into a real thing. The demand was real and the general large-scale investment was justified.
Just because some companies made bad decisions doesn't mean the railroad industry as a whole was some kind of mirage or mistake. Laying down tracks for trains and passengers that didn't exist yet is still necessary.
>Except the "profit that is mathematically impossible" part. That's just made up and false. It's entirely possible that we are actually underestimating demand, and there is going to be tons of profit.
JP Morgan says $650 billion in annual revenue required to deliver mere 10% return on AI buildout is equivalent to $35 payment from every iPhone user, or $180 from every Netflix subscriber 'in perpetuity.'
Very, very, very unlikely it makes profit, which why AI keeps getting overhyped by CEOs.
Altman said months ago that they are expecting around $65/user/month from ad-supported ChatGPT. A strong hint about where they see account prices in the future.
When you run the numbers, $65/mo turns AI investment into a a 5-7 ROI, which is totally within normal bounds.
Considering there are over a billion unique weekly active users for the major labs, and demand has been relentless, it's a pretty easy sell to get investors on board.
Those numbers sound... unrealistic to me. Just doing some napkin math: 65 $/user/month / 0.01 $/ad ~= 6500 ads/user/month, which is about an ad per minute if you assume someone is using the chat interface 4 hours a day including weekends. Maybe you see that behavior from "my GF/BF is AI" types but I'm also already assuming 0.01 $/ad which is super high to my understanding (if you work in adtech please correct me if I'm wrong). I don't forsee over 50% of your workday or leisure time spent in ChatGPT as likely, especially if the ad rate is well beyond YouTube's nigh-unusable amount it is now.
Did they consider that profits on the build out won't be uniform, i. e. there will be some companies that go under but the rest of them will capture the profit?
Some companies going under doesn't change anything about the market as a whole.
If the demand is real and the company just sucked, their users and infrastructure will end up at a competitor: the value for that one company is bigger, but the overall per-user bill remains about the same.
If the demand is fake the infrastructure will be sold off at a big loss, allowing new companies to enter the market with far smaller investment costs, allowing them to undercut the competition, driving down the price users expect to pay for compute, resulting in a race to the bottom between the remaining AI companies in an attempt to attract enough users that their hardware won't sit idle - which in turn makes it far less likely that they'll be able to hit those revenue figures. And a bunch of investors just lost a few billion dollars, of course.
Consider that a large majority of the revenue will come from businesses.
Even $100 per month per employee will likely turn out to be quite reasonable, if it can make employees more productive by several hundred dollars per month.
the big question is: who would pay for those services then?
I mean I love simplicity, and if economy could be simplified to a big money printer machine directly printing the money to a burner, then it would be so simple, that even a short context window could comprehend the economic cycle finally!
LLM's can actually be exceptionally good at research and pattern recognition, i.e. analysis. And while they aren't great at running numbers themselves, they can do exceptional work passing off Python scripts to an interpreter to generate the numerical results they need.
I'm quite sure the Robinhood AI is going to be trash, i.e. just a gimmick.
But, it's not crazy to think that with the right harness, there are big opportunities for identifying profitable strategies. Especially relying on unparalleled and essentially unlimited research capacity based on public information. More analysis than any single firm could ever hire.
And even for Robinhood users, it's entirely plausible that AI-traded stocks will perform much better than the trades a majority of users would make, since most investors are really unsophisticated.
reply