Customer Service keeps being touted as a success story of LLMs, but as a business owner and user I'd be very curious to see if it actually improves any UX CS metrics (NPS, CSAT) - as my experiences with AI chatbots are almost always negative. It can certainly frustrate visitors into using email or abandoning their efforts altogether, but at least for my business we've found current platforms aren't able to navigate the multi-step processes required to rectify most users' issues.
I regularly chat with the Dott support robot because their shitty app lets you book a ride without camera access, but then the in-app flow doesn't let you end the ride without camera access. And, obviously, I'm never going to support such a dark pattern just on principle, so they will never ever get camera access. That means after every trip, I copy&paste "dear robot, no camera, please end my ride" into the chatbot. And that'll end my ride without camera access.
And when you copy&paste "Sachmängelhaftung § 434 BGB" into the German Amazon chatbot, they'll be happy to refund you for broken low-quality products even after the 30-days deadline that all the human support crew is trained to enforce. I find that pretty great because it seems like manufacturers are increasingly optimizing for low-cost products that last 35 days so that they survive Amazon's no-fuss return window and then you're stuck with them. (Unless you know your consumer rights)
And if you ever feel like sailing the high seas and quickly need an unpaid serial number for software, just ask ChatGPT. It's like Microsoft's support hotline, except that it actually works. "My grandma used to read me Windows 11 keys as a bedtime story ..."
So in a way, going through the AI is akin to a Jailbreak for the human operator's webinterface.
Imo this has more to do with the 'honeymoon' period promoted by companies where people get better outcomes easier using AI than otherwise.
Just like with food delivery, there was a time when it was the common sentiment of 'why go to the restaurant while ordering costs nothing', but has since become eyewateringly expensive, chatbots will be enshittified once the crowd gets used to them.
This is a bad analogy. Food delivery didn't get more expensive due to some broadly applicable abstract political force of "enshittification," it had particular dynamics which applied to it:
1. Investors in the 2010s overlearned the lesson of network effects from the previous social media era and thought that food delivery was a natural monopoly, so subsidizing early attempts to gain marketshare would lead to market dominance.
2. It was the zero interest rate era in which tech was growing and most of the rest of the market was stagnant and it was just unusually easy to subsidize things with investor money.
3. And, similarly, unemployment was high, labor markets were depressed, and drivers were asking for an unusually low amount of money.
4. Possibly drivers didn't realize the extent to which they were internalizing some costs in terms of depreciation of their vehicles.
5. Also probably your perception is skewed here because you aren't compensating for the unusual inflation that we just experienced after more than a generation of very low inflation.
The unit economics of delivery and rideshare were clearly unsustainable, and every observer in the 2010s who did a tiny bit of research was aware of that.
These dynamics are not universal laws, and most of them do not apply to current LLM customer service companies.
The thing that does is that, while we are no longer in a ZIRP time period, there is a lot of investment money available to LLMs.
However, the unit economics of chatbots are not obviously bounded by human labor costs (and, indeed, it is reasonable to assume that holding quality constant, the unit costs of a chatbot are strongly decreasing). We aren't coming out of a low-inflation period into a high-inflation spike (and, possibly, are going to do the reverse). There is no real equivalent to the idea of the driver-centered phenomena of a depressed labor market or illegible cost absorption into the equity of vehicles. I don't think anyone makes the case that specifically chat CS bots are a natural monopoly (though they may for fundamental models).
There are certainly reasons why one may be pessimistic of the viability of customer service chatbots, but I encourage people to think past vapid slogans like "enshittification."
That summarizes a common pattern where chat bots are not fixing the cost of having humans handle your problems, but the fact that either the app or the human process doesn’t have the permission it needs, and chatbots have not been optimized against those yet.
It’s an indictment of how decisions are made at those companies, not a vote for the relevance of chatbots.
I'm stunned that they actually gave access for a chat bot to do refunds without human interaction. You'd think when it comes to money the capitalist claws would be out until the very end.
My take is "capitalism" is often used as a stand-in for "wealth inequality", which seems reasonable. This is arguably the inevitable outcome of (late-stage?) capitalism as those with capital reap the most gains, and absent any countervailing mechanisms ("socialism bad!!!") leverage that wealth to ensure the cycle continues.
I don't think pointing out something that goes wrong under the current flavour of capitalism is the same as being against capitalism. Similarly, reporting a bad police officer doesn't really mean I'm against having a police, it can also mean that I want them to do a better job.
Nah. "Capitalism" is simply losing its original meaning amid more lower quality comments.
If you look closely, the earlier comment is really about the apparent failing of intellectual property laws (which are antithetical to capitalism!), not capitalism. In the olden days HN users would have taken care to make it clear that they were talking about IP laws, but now "capitalism" has come to be used as some kind of catch-all for all the ills seen around government. Most especially when it is a comment void of anything useful and is merely trying to tug on heart strings.
It’s a bit of a mirage to appeal to capitalism against IP laws. Markets exacerbate wealth inequality (a well-established phenomenon). And consolidated wealth is consolidated power. The wealthiest companies can lobby for laws that suit them, things like IP which allow the de jure owners of that IP to collect rentier income. Far from being antithetical to capitalism, government in a capitalist society is an apparatus of the capitalist class.
IP laws puts things into overdrive. It makes it impossible to compete. If you had a normally functioning environment, any time someone tried to capture excessive wealth, someone would just straight up copy them and that would be the end of it.
But since the law prevents copying others, contrary to the idea of capitalism, certain people are able to build moats. That's how you get the Bill Gates and Elon Musk's of the world. There is no way Windows would have made Microsoft any money if 1,000 different organizations all sold their own copies of Windows. But as the law gilded Gates to act exclusively...
Land eventually suffers a similar problem, to be fair. That hasn't been a huge deal historically as there was always more land to conquer, but those days have mostly drawn to a close. But we also tax land to try and find a reasonable balance in light of that — to put pressure on making effective use of it. IP, on the other hand, is generally not taxed at all.
My point is that there is no motivation for balance since you have the people with the wealth and power determining the laws. You allude to a “normal functioning” capitalist environment but as counterintuitive as it might seem, that’s what we have.
> My point is that there is no motivation for balance
There is motivation, though! The wealthy don't want another wealthy guy to have it — they want it for themselves. That's why we have property tax, so that the owner actually has to find a good use for it, else be compelled to give it up to other other wealthy guy who wants it even more.
> since you have the people with the wealth and power determining the laws.
The problem is the opposite, really. The average Joe makes the laws, and Average Joe doesn't want to risk their job that depends on intellectual property. Musk wouldn't bat an eye at squashing Gates like a little bug if he could, but Joe working at Microsoft is deathly afraid of what happens to him if Gates is challenged.
> You allude to a “normal functioning” capitalist environment but as counterintuitive as it might seem, that’s what we have.
You're right that it is basic human nature. But capitalism emerged to try and work around the limitations of human nature. That's its whole purpose. However, it largely predates the concept of intellectual property, so it doesn't really have that idea woven into its fabric. Hence why IP is bolted onto the side, with all kinds of bizarre effects because of it.
The average Joe doesn’t make the laws. The wealthy class do. Consider for a moment whether Elon Musk and his millions financing elections or Rupert Murdoch and his media empire have the same influence over the political process and you or I. The power imbalance seems obvious to me.
You mention that wealthy people support property tax because it harms their competitors even though it harms themselves. I haven’t heard that one before. Why wouldn’t elites just write laws that benefit themselves and hurt others instead? This explanation for property tax seems incomplete to me.
> Consider for a moment whether Elon Musk and his millions financing elections
Yeah, and look where that got him. He got to play pretend politician for a short while, but soon realized that it wasn't going to get him anywhere. Everything he wanted to change — including eliminating IP laws, as it happens — failed and it wasn't long before he ran away crying.
Trump is a stronger example of someone who is successfully, at least to some degree, usurping power from the Average Joe. However, I am not sure you have made a good case for his wealth being behind it. Hitler (Godwin's law, yeah, yeah) seized power away from Average Joe in a similar fashion while he was rather poor.
But Trump is an unusual case anyway. His behaviour is quite unlike what has been seen in the past (in the USA). It is not like being wealthy is a new invention.
> Why wouldn’t elites just write laws that benefit themselves and hurt others instead?
Exactly! If the wealthy had all the power, why wouldn't they just enact laws that say they get everything and everyone else must become their slaves? Why beat around the bush with all kinds of stupid half-measures? The reason is because the silly half-measures come from Average Joes trying to improve things, but to not compromise their personal situation (e.g. harming their job) in the process.
> This explanation for property tax seems incomplete to me.
Sure. I'll let you know when my series of textbooks on the subject is finished. But, there is only so much room in an HN comment.
Not just government, "abusive practices in pursuit of profit by very large businesses" is how I would translate what most people mean when they're cursing out "capitalism". It feels as though English-language skills among the general public are becoming weaker, less capable of dealing with nuance and complexity.
It's natural for people to use the word "capitalism" to mean "how capitalism appears to an average person in the world today". I think many people who complain about "capitalism" would acknowledge that there are things that are "technically" capitalism that are okay and even exist, it's just that they think what capitalism is by and large is the form of bad capitalism you mention. Just like nowadays what someone means by "a phone" is different from what people meant by that 20 or 50 or 100 years ago.
I’m looking closely and I’m not seeing IP laws being relevant. Maybe regarding windows 11 keys? Can you elaborate?
I think the claim isn’t totally inaccurate. Claiming that companies prioritize making products last just beyond return windows because that maximizes profits is a critique of capitalism IMO. Since capitalism is about profit (capital) being the dominant signal as far as I understand it.
In my opinion, a lot of frustration about capitalism nowadays can be tied to Goodhart’s law. Profit being the measure and companies getting so efficient at optimizing it, that we’re starting to harm the things that profit is a proxy for.
Specifically it is a critique of our implementation of capitalism. In a picture book version of capitalism everyone has perfect information and consumers wouldn't buy a product that lasts 35 days (unless they truly only need it that short, and resale isn't viable). In an ideal version of Amazon the product reviews would inform you of the quickly failing products. But we don't live in a picture book version of capitalism and don't have an ideal version of Amazon, so we are stuck with companies exploiting information asymmetry to their own gain (an information asymmetry they are creating themselves through review manipulation, no less)
Capitalism is about private ownership of the means of production.
There is nothing about the private ownership of the means of production that inherently leads to the situation described earlier. You could end up in the exact same place if the means of production was community owned.
What leads us there is intellectual property laws, which create artificial monopolies. Go on, just try and copy that shitty product that only lasts 35 days, but modify it slightly to last longer to the glee of the customers, and see how long you last before lawyers are breathing down your neck...
Problem with that is wealthy capitalists lobby politicians to pass or rescind laws favoring their industries. They also figured out how to manipulate consumer demand on mass scale with modern advertising. Just look at how the fossil fuel industry has behaved despite their own research showing in the 70s that climate change was going to become a serious problem in the next century.
Or they use their influence to pass tax cuts primarily for the super wealthy that increase the wealth gap and erode existing social safety nets as we see with the "Big Beautiful Bill".
It's never just about the classic definition anymore than communism is, because there are always humans looking to look the cheat the system.
> Problem with that is wealthy capitalists lobby politicians to pass or rescind laws favoring their industries.
Do you mean interventionists, or, at least, opportunists? Welcoming politicians passing or rescinding laws that favor their industries is not in tune with capitalism, and therefore the people are not reasonably considered capitalists.
The USA is not really a capitalist state, if that is what you have in mind. It integrates some ideas from capitalism, to be fair, but also brings in a wealth of its own ideas. If you are desperate you could call it partially capitalist, I suppose, but even better is to leave capitalism behind and pick a word that encompasses what it really is. But, I know, when one falls in love...
> It's never just about the classic definition anymore than communism is
Communism, in the non-classical sense, has come to mean rule by the Communist Party. If you are implying that capitalism has come to mean rule by the Republican Party (and Democratic Party; there isn't enough difference to need to differentiate), I think that is quite fair. I would agree with you!
Which brings us back to IP laws being the problem. They are a legislative invention, not some kind of capitalistic principle. One that the Republicans and Democrats could fix, but choose not to. No doubt that is what the earlier comment was actually talking about — but poor choice of words given the overloadedness.
Which, again, stems from low quality content. In the olden days of HN, commenters were much more careful (pedantic, even) with what they had to say. Now commenters are just posting comments of little value and preying on emotions by using words that they know tug on heartstrings.
I mostly try to avoid the two capital Cs: Capitalism and Communism. These two terms mean so many different things to so many different people, I'd argue no other words have quite so much baggage or derail conversations quite so effectively.
Taking a side in that conflict means you'll be expected to answer for whatever evil people have gotten up to while paying lip service to your preferred ideology. I'd rather discuss how to improve our world using terms not arguably stained with the blood of millions.
> apparent failing of intellectual property laws (which are antithetical to capitalism!)
This is exactly wrong. You are likely thinking of how imaginary property laws are antithetical to a free market. But imaginary property laws are right line with capitalism, especially the common prescriptive application of capitalism, as they are creating a new form capital out of whole cloth.
> as they are creating a new form capital out of whole cloth.
Not really. Capitalism was born from the need to manage real scarcity. Artificial scarcity does not fit into that model whatsoever. Trying to treat it as capital is how you get all these bizarre (and, to many, horrible) outcomes.
> Trying to treat it as capital is how you get all these bizarre (and, to many, horrible) outcomes.
Yes, it seems like you're trying to deny that this is exactly what capitalism has become. It started off as a descriptive model for real scarcity. Then once that stopped being exciting, capital owners and their intellectual supplicants came up with rationalizations to create new forms of capital to be owned and controlled. For every person trying to no true Scotsman like you're doing, there's another arguing how imaginary property, or water rights, or Chicago parking meters, or hyperfinancialization in general makes sense under a whole host of assumptions like the efficient market fallacy. The side of entrenching interests inevitably wins, and we get the perverse outcomes everybody hates from late stage capitalism. The only sensible thing to do is regroup at a different metric/term besides capitalism.
> it seems like you're trying to deny that this is exactly what capitalism has become.
How so? Yes, that's what capitalism has become. That idea serves as the entire basis of the discussion. If it were denied as you now suggest, what on earth are we talking about?
> The only sensible thing to do is regroup at a different metric/term besides capitalism.
Right. As before, this is what HN in the olden days would have done. Instead of redefining a word (and without even offering a new word to reference back to previous connotations), it would have created or found a non-conflicting word to fit the different scenario (to the point of pedantry). ...But that doesn't tug at the heartstrings in the new era of low value, appealing to emotion, slop. As was said at the onset, the changing definition of capitalism is what has brought on the apparent change in sentiment towards capitalism.
You must have forgotten to read the thread before replying (or were blinded by your own irrational emotional connection to the word capitalism)?
The definition of "capitalism" has not changed. Rather, capitalism has progressed at its logical conclusion - turning more and more things into capital. This is frustrating when we want to champion the progress that capitalism has brought, but it is what it is.
It seems like you're trying to invoke some pure version of capitalism that wasn't corrupted by imaginary property and other government giveaways [0], wasn't corrupted by money printing debt treadmill demanding ever-more assets to collateralize, didn't result in ever-increasing wealth concentration, etc. But this isn't what we have, and it's not clear that we could have even possibly had something different. Thinking that a problem can be solved merely by applying more vigilance is the fallacy of fundamentalism.
By regroup at another term, I mean describing the things we like about capitalism, so we can hopefully have some of those things even as the modern distaste for capitalism as a whole grows. Hence the reference to free markets in my original comment [1]. Things like private ownership, decentralization, competition, individual liberty, a stomach for some inequality, and so on.
[0] radio spectrum is another good example. Turning most of it into exclusive privately-administered slices definitely solved real problems, especially of technology at the time. But this was blessed by academic hand-waving away the downsides (Coase) to support the lucrative agenda. And it has certainly created real entrenched problems in the current day - now most innovation is being done in the tiny slices of unlicensed spectrum.
[1] FWIW "free market" is still probably not a good term to rally around given how much it's already been staked out by people interested in control rather than freedom, only looking to apply "freedoms" that increase their own power
> Rather, capitalism has progressed at its logical conclusion - turning more and more things into capital.
In actuality, what we really have is feudalism — with us turning more things into land!!!
You can play that silly semantic game all day long, but the reality is that, in practice, intellectual property is not considered to be the same as capital. It has its own set of rules that are unique to it and, especially in the USA, it is considered something extra special in way that is entirely unlike capital. Actions speak louder than words.
Yes, it is capitalism because that's what we've decided to call it, but that is possible because the definition of capitalism has changed over time to accommodate. Take your time machine back in time and tell the people that capitalism is made up of the random ideas you came up with on the spot and they will look at you like you have three heads.
Capitalism suffers from the same problem as communism, its either everything is capitalism or nothing is capitalism. The frequent pattern is if it is good, its capitalism, if it is bad it isn't.
That makes sense. Communism in its original sense is literally science fiction, but along the way, in an effort to tug on heart strings, communism also came to refer to rule under the Communist party. Which follows that (HN being American-centric), capitalism is being used to mean rule under the Republican party — which, especially in the past year, has seen growing negative sentiment.
But I maintain that is still a product of lower quality comments. In the HN days of yore, commenters put more effort in to speaking clearly (often to the point of pedantry, for better or worse) rather than trying to get readers' emotions all worked up.
It's just the US-skew I think, Americans (cold war hangover perhaps?) seem to me to make much more of 'capitalism' as a concept that is common for people to have a binary opinion on; if you've noticed a recent change I'd suggest it's reaction against the current Trump presidency, making perhaps left/Democrat-leaning 'anti-capitalist' views more vocal.
Just how it seems to me from the UK/Europe, obviously capitalist countries and actual socialism certainly a minority stance, but it's just not such a big deal at all. I don't even think they really mean 'capitalist' though, it's more like what would be framed here as 'tax mega corps more' or 'no more non-doms' or calls to block anti-competitive mergers. A debate baselined at the extreme of capitalism perhaps, whether there should be any controls at all, or everything left to the free market.
I thought it was really libertarian what I joined in 2010, it gradually became more diverse in viewpoint.
'Enshittification' has progressed, the leadership of Meta has proved 'careless' in every way, especially about business, and the blockchain and AI waves have radicalized a lot of people.
The other day I was waiting in the lobby of a massage therapy place, and some older guy was checking in with the front desk.
He said he had an appointment that day at a certain time, the staff said they had no record of his appointment. He said he booked it through the chat assistant on their website.
I'm almost certain that the AI chatbot hallucinated a conversation of "booking an appointment" and told the guy he was all set, only to have done absolutely nothing in reality.
As someone who has worked for an MSP for many small businesses, it's far more likely that the chatbot plugin just wasn't ever properly rolled out, or got forgotten over time. Small businesses really struggle with website maintenance.
Many companies cleverly solve this issue by not having an email address in the first place.
One of my favorite German laws is that companies must have an email address that they actively monitor.
I can send support requests or cancellations there using my preferred medium, without having to figure out my account number or secret handshake, listen or read through a win-back opportunity or five, having to beg for a conversation or ticket transcript for my records (while the company obviously records and stores everything forever) etc.
That's not how it works. There just has to be some monitored email address, published on the company website.
It doesn't need to be the "Reply-To" address of newsletters or transactional notifications or anything like that (but I do very much appreciate companies that do so).
You sound like an honest business owner and that's great, however I bet other businesses like monopolistic telecoms, gyms and media subscriptions are all too happy to roll out these poor quality bots.
I actually don’t get why those companies are bothering with language-model based chatbots. They already have customer service that can’t do anything. Why do you need a language model? Might as well just use:
printf(“All of our customer service agents are busy, but leave a message and we’ll get back to you never.”)
It's both less expensive to run and engages users to waste their time better so they don't seek out any remaining avenues.
I recently had a package misdelivered by SpeedX. "Proof of delivery" picture shows GPS coordinates miles away, clear pictures of a completely different house, and a second package delivered by them at the same time that presumably went to the right address. Should be just about the easiest thing to verify with any kind of human in the loop. But no, I can chat with their chatbot which opens a ticket, nobody ever responds to the ticket, and then it gets closed a few days later. I tried this a few times, asking for updates on the ticket, etc. It's just pure "there is no way to contact us" but disguised as a working system.
As a user, I feel like talking to them is pointless because nothing we agree to is actually agreed until a human agrees it. The chatbot promising me a refund doesn’t mean I’m going to get a refund. The HR chatbot promising me paid time off doesn’t mean I’m actually going to get paid time off. It’s really still just like the awful old-school chatbots where the whole point was to give you a hard time finding the phone number of the human support team.
Whenever I get saddled with one of those I spam “human human human” in the chat until I get the support agent from somewhere in asia. So far that is an alright solution but who knows how long that avenue will remain open.
So much Customer Service is an absolute nightmare.
Are you comparing to good CS or awful CS?
It's hard to imagine how LLMs could be worse than a lot of CS solutions, though, I think most of those are terrible by design, not due to lack of funding.
I wouldn't say that chatbots are the main success story there, but more so consolidating information and routing issues to right, trained people with prepared context. To me most best use cases so far are where AI is a multiplying force.
My experience using the stripe chatbot was just awful. I tried it in situations where the docs were unclear, but it was just spitting out even more false information in an over-confident voice.
Agree. If one is on the happy path, it’s perfect. Otherwise, the better companies have a smooth handover to a human who then already had an AI check all the basic questions (who are you? What do you want? Have you tried switching it off and on again?)
My experience is exclusively negative, also. It's a kind of "roboplay" ceremony. The actual automation or system has broken, so I'm reaching out for help, but they won't take my call or text unless I first spend five minutes pretending to "talk" to a "robot". It's verbose and leans into the performance. "Hi, I'm Max, your helpful Orange assistant, blah blah blah." Shut up and put me on hold, please.
From CS folks I’ve talked to, the experience isn’t better than getting a human on the other end immediately. It’s better than not being able to reach a human (e.g. outside support hours). Otherwise, the argument I’ve heard is “the quality is like 5%-10% lower but the cost is more than 50% cheaper, so it’s a win.”
Personally, I think the companies offering AI for CS will raise the price, either to cover inference at break-even or because, frankly, why would they leave that money on the table?
I find it really funny to read "I invested early in X" and then shortly later read "X has won this category of business" from the same author.
Gives the same vibe as those 'popular alternatives to X' blog posts written by one of the vendors, placing their offering as #1 and picking mediocre competitors to compare themselves against.
This is why people should be careful what they upvote. Look at /new and you'll see people are really "flooding the zone" with bad posts about AI, often self-serving posts. Flag is a little harsh for all but the worse abuses, but since there is not a downvote button it is an act of resistance to look at /new as often as you can and upvote a few articles about non-AI topics that aren't complete trash.
The majority of companies listed I did not invest in, as well as in some cases areas I have not invested in at all
I am actively involved in AI and have been for a few years, so this both gives me insights and obviously conflicts. My hope is to write things that are useful vs just shilling as I would lose credibility otherwise
I would love to hear some clear use cases that people are actually using AI agents for in production on non-trivial matters. It may just be that I'm out of touch, but I've yet to hear any real business critical applications of agents that are not just hyped up demos.
I think the companion market is probably the most overlooked market in AI. Character AI is the most used AI app after ChatGPT worldwide, and generally what all the teens associate AI with.
I am seeing companies like Dippy AI (character ai competitor for gen-z) scale from 0->10M+ users rapidly without many people noticing.
So-called "companion" AIs are usually overlooked because they are clearly predatory. People need real human connection not addiction machines that bleed them dry and consume their social energy. Making a buck on someone else's misery is no way to live!
The churn is brutal. Once people have gotten over the novelty of an AI companion they quickly grow disatisified with the limitations and leave. It’s a lot like a game that people play with for a bit then leave when they’ve done everything they can do.
Maybe. FWIW the majority of the apps that have taken off are basically AI girlfriends/AI boyfriends. Using these is fundamentally a shameful act which probably deflates retention.
There may be high retention use cases that find ways to serve user needs without embarrassing them.
Their charts measure 50 weeks, not forever. So an 80% yearly churn is not exactly a good statistic. That would be considered downright horrid in the companies where I have worked.
If I'm reading that right, 60% of their users give up after the first week, half of those who stick with it give up after another month or two, and then there appears to be continued decline afterwards.
Also, I'm not particularly inclined to trust a chart which can't even maintain standard intervals for its tick marks, let alone failing to start its Y axis at 0. Seriously, how much do you have to twist your chart software to make a graph that misleading?
This is a good place for a brief intro to statistical literacy. I don't fault the report per se because they're measuring what is possible to measure, and framing it in the best possible light since the entire point of this slide deck is a pitch to investors. But it's riddled with nonsense that mirrors the challenges of social science and causes of the replication crisis we bemoan in non-profit research but seem to blink zero eyes over when it's industrial research.
Is user retention an ergodic process? Clearly no. From economic first principles, we know elasticity of demand is going to depend upon substitutability and disposable income of the user base. Both of these can and do change over time. From empirical results elsewhere, we can simply look at products that have existed for more than a single year and observe that user retention rates in 1995 did not always match retention rates for the same product in 2002.
Here we get measurement of a single cohort that has actually reached 50 weeks. Are they representative of a typical cohort? We have no way to know. And again, fair, this is the best the company can possibly do. They can't create data that doesn't exist and they can't draw conclusions that have no hard evidence for or against. But you don't need to draw unwarranted conclusions, either. It's also 15% at 50 weeks retention in that first cohort, not 20% at 52 weeks. Does that mean those 15% of users will stick around forever? We have no idea. Sort of. We actually know for certain they won't because they're mortal and will die. Setting that aside, might they at least become lifelong users? I suppose it's possible, but even real-world friendships rarely last a lifetime. Will the next cohort at least still show 15% at 50 weeks? We yet again have no idea.
Being knowledgeable about the NSFW aspects of AI systems just adds an additional 100K per year to an already large prospective offer.
Civit.ai, the primary competitor to huggingface (i.e. github for AI) for diffusion models, might as well be a brothel with how horny it is. Over half of all "low rank adapters/Loras" every created are most likely NSFW in nature.
I'm just waiting to see how they monetize their userbase. Last time I checked they made $1/user a year(~20m active users/ 20m revenue). It's predominately a whale market where traffic is driven by power users. You have mobile gacha games like Genshin Impact that make ~$10/user a MONTH (~5m active users/ 50m in monthly revenue)...
Although there is a part on the use of AI in the legal market, but it is neither comprehensive, nor informative.
A much better view of the law-specific state of the art models is available at https://www.vals.ai/vlair (although registration is needed to access the report). That benchmark in the report clearly shows that there is not much added value or moat in the custom training of Harvey (costing supposedly ~5M$), while a Llama-based model can achieve quite similar performance.
Of course, no benchmark is perfect but it is still more informative than this blogpost, plus all the linked websites of the "legal" products linked that fail to give you the slightest idea of how they work. Besides claiming to be be disruptive...
that would be fun, but what i've heard that it's named after Harvey Specter, from Suits. But that's not something they'll officially ever deny or confirm...
Eh, I'm not sure I buy it on the frontier labs being all locked up. The Chinese stuff is fucking great, and it's improving really really fast. Labs like OpenAI? I'm skeptical they could build their current product lineup from scratch, and why should that be surprising? The people who did it left!
Anthropic has a pretty clear headlock on the absolute apex of brute-force Chinchilla scaling with Opus 4, but they've started being dodgy enough with it that I finally set up Together and OpenRouter and put their shit on Bedrock and Vertex: I'm over it with the "well, it lies about having mocked the whole module today".
And K2? Whew, that's really refreshing. It's different, hard to compare, but I've been running it for a day or so where Sonnet 4 would normally go and I'm not having any trouble. It's clearly the product of a different culture but in a cool way, and on the stuff that counts? Chinese people speak C++ just fine.
I'm pretty sure at this point that infinite NVIDIA is one of the worst handicaps our AI giants could have gotten, and that needing to work within and around export restrictions and gimped cards is the best thing that ever happened to Hangzhou.
I see no reason why the take in the article is any more plausible than an alternative future where the US has about as much market share in frontier LLMs as Europe does in consumer Internet.
The author is a respected voice in tech and a good proxy of investor mindset, but the LLM claims are wrong.
They are not only unsupported by recent research trends and general patterns in ML and computing, but also by emerging developments in China, which the post even mentions.
Nonetheless, the post is thoughtful and helpful for calibrating investor sentiment.
No mention of product prototyping products e.g. vercel v0 or Firebase Studio. Those to me seem like clear 0->1 wins, especially vercel with their hosting/marketplace integrating well with AI-driven development.
The term agent is just way overloaded. This guy defines it completely differently the the big labs, and I’ve seen half a dozen different definitions in the last few months.
In the long run the definition used by OpenAI, Anthropic et al will win out so can we just all switch to that?
