Claude’s memory architecture is the opposite of ChatGPT’s
shloked.com432 points by shloked 2 days ago
432 points by shloked 2 days ago
The difference is implementation comes down to business goals more than anything.
There is a clear directionality for ChatGPT. At some point they will monetize by ads and affiliate links. Their memory implementation is aimed at creating a user profile.
Claude's memory implementation feels more oriented towards the long term goal of accessing abstractions and past interactions. It's very close to how humans access memories, albeit with a search feature. (they have not implemented it yet afaik), there is a clear path where they leverage their current implementation w RL posttraining such that claude "remembers" the mistakes you pointed out last time. It can in future iterations derive abstractions from a given conversation (eg: "user asked me to make xyz changes on this task last time, maybe the agent can proactively do it or this was the process last time the agent did it").
At the most basic level, ChatGPT wants to remember you as a person, while Claude cares about how your previous interactions were.
The elephant in the room is that AGI doesn't need ads to make revenue but a new Google does. The words aren't matching with the actions.
The bigger elephant in the room is that LLMs will never be AGI, even by the purely economic definition many LLM companies use.
I've been saying this for years now. LLMs are _not_ the right methodology to get to AGI. My friends who were drinking the kool-aid are only recently coming around to "hey, this might not get us AGI".
But sometimes it feels like I'm the lone voice in a bubble where people are convinced AGI is just around the corner.
I'm wondering if it's because people are susceptible to the marketing, or are just doing some type of 'wishful thinking' - as some seem genuinely interested in AGI.
Yeah I've had those conversations since GPT-3 first came out. I usually look like the one way off base, but I never did hear a clear explanation of how the LLM architecture could lead to AGI.
In my experience it was a combination of the hype and an overconfidence in the person's understanding of how LLMs work and what AGI actually means. To be fair, AGI definitions are all over the place and LLMs were rarely described in detail beyond "its AI that read the whole internet and sounds like a human."
There are two big innovations required to achieve inexpensive AGI.
LLMs will accelerate discovery and development of Innovation 1, for insanely expensive AGI.
Innovation 1 will accelerate discovery and development of Innovation 2 which will make it too cheap to meter.
> LLMs will accelerate discovery and development of Innovation 1, for insanely expensive AGI.
Can you expand on this more? As far as I'm aware LLMs have yet to invent anything novel.
At best they may have inferred one response of many that, when tested by humans, may have proven out. I'm not aware of a specific example of even that, but it is at least possible where claims that LLMs will "cure cancer" seem plainly false (I'm not trying to put those words in your mouth, just using an example for my point).
To reword the downvoted sibling commenter's intended point:
> The elephant in the room is that AGI doesn't need ads to make revenue
It may not need ads to make revenue, but does it need ads to make profit?
Don't fool yourself into thinking Anthropic won't be serving up personalized ads too.
Anthropic seems to want to make you buy a subscription, not show you ads.
ChatGPT seems to be more popular to those who don't want to pay, and they are therefore more likely to rely on ads.
In the 2020s, subscriptions don't preclude showing ads. Companies will milk money in as many ways as they can
(Netflix as an example)
And cable companies, and magazines. This is not something from the 2020s, it is a centuries old thing.
But these are entertainment. For all the time advertising has been present, work tools have been relatively immune. I don't remember seeing ads in IDE for instance, and while magazines had ads, technical documents didn't. I have never seen electronic components datasheets pitching for measuring equipment and soldering irons for instance.
That's why I don't expect Anthropic to go with ads it they follow the path they seem to have taken, like coding agents. People using these tools are likely to react very badly to ads, if there is some space to put ads in the first place, and these are also the kind of people who can spend $100/month on a subscription, way more than what ads will get you.
They might be coming from different directions. But these things, as often they do, will converge. Too big of a market to leave.
and netflix used to think they dont want to show ads either.
Netflix likely doesn't want to show ads, but the market would rather watch ads than pay full price for a service.
https://www.theverge.com/news/667042/netflix-ad-supported-ti...
> the market would rather watch ads
no, netflix wants more income, and by having a product be ad supported, they can try to earn more.
The "market" is not a person, and doesn't have "wants".
From the article:
> Netflix has more than doubled the number of people watching its ad-supported tier over the last year. At its upfront presentation for advertisers on Wednesday, the company revealed that the $7.99 per month plan now reaches more than 94 million users around the world each month – a big increase from the 40 million it reported in May 2024 and the 70 million it revealed last November.
1/3 of Netflix users (the market) prefer ads over paying to avoid them.
A lot of "netflix users" are middle and high school age kids in third world countries using a borrowed account. User context matters a lot. If someone's friends-friends-friends uncle changes their password, it's no surprise those "netflix users" would switch to an ad-supported model. It's possible but unlikely the 12 year old kid watching anime on a shared/borderline stolen account has the resources necessary to buy an ad free account at US prices.
But the ad-supported tier isn't free either.
I don't think the difference for a 12yo is $7.99 for standard with ads vs $17.99 for standard.
It's $0 vs any non-zero dollar amount.
This leaves me somewhere between surprised and shocked.
Maybe you shouldn't be. The ad-hating paranoid HN user is not representative of the general population. Probably the exact opposite, in fact.
My wife and mother love ads, they are always on the hunt for the latest good deals and love discount shopping. When I tried to remove the ads on their computers or in the postal mail, they protested. I think they are far more representative of the general population.