jsw97

6 ms·

1.
by jsw97 3d ago
I know this is intended as a joke but there is something to it. The public market has a way of elevating profit over everything else, even if you start with the best intentions. Remember “don’t be evil”?
2.
by jsw97 6d ago
One complication for geofencing is that the Mass law applies if one party is in the state. So if you're on a zoom with someone in Mass, you're still violating.
3.
by jsw97 6d ago
Just to be clear, I have not seen this behavior. If this is true, though, then given the way their chat operates, this might be more dangerous than it seems. One of the things I like about ChatGPT is its memory, the way it kind of seamlessl
4.
by jsw97 6d ago
Even beyond this, if you create circumstances where mass commission of consequential felonies is an inevitable result, I suspect that is also illegal. 100% this just gets ignored by the Mass attorney general though.
5.
by jsw97 6d ago
Not sure why my comment merited a downvote. In any case, quoting from Mass law: https://malegislature.gov/Laws/GeneralLaws/PartIV/TitleI/Cha... The term ''interception'' means t
6.
by jsw97 6d ago
Argument for only wins because of the weight of the lawyers they can bring to bear. Clearly ephemeral recording for the purposes of processing violates the spirit of the statute at least. And there is this, from the MGL: The term &#x2
7.
by jsw97 6d ago
This is, I believe, a felony in Massachusetts. https://www.romanolaw.com/can-i-record-a-conversation-in-mas... (Linking to discussion, not the statute, because it confirms that it is in fact enforced.) Come to think of it,
8.
by jsw97 7d ago
Even you think both cheat, you can't possibly think they both cheat the same amount.
9.
by jsw97 8d ago
The last time I checked there was a loophole — if you provide feedback in-session (responding to “how are we doing” or “which prompt is better”) then they can use that feedback + relevant context. Relevant context for chatgpt might include
10.
by jsw97 8d ago
Yeah this is a land mine. Even if you opt out of them training on your conversations, giving “feedback on a new version” or answering “how are we doing” or whatever can slurp up all relevant context, which if you think about it can probably
11.
by jsw97 8d ago
I think it’s probably not surprising that they would go up to the contractual limit or into a grey area; but exceeding that would require too much coordination among individuals, as you say.
12.
by jsw97 8d ago
Would that be more or less unlikely than accidentally hacking another company? More or less unlikely than colonizing an obscure wiki? Highly persistent agents + vibe-coded security seems like a problem.
13.
by jsw97 9d ago
A blog or YouTube channel would be a nice halfway step to test the waters. A publisher might like to see that anyway. It would give you a natural feel for what the appetite is and where you want to put the level of the presentation.
14.
by jsw97 12d ago
Sure they make sense — you need some incentive to drive the price to zero.
15.
by jsw97 12d ago
In the past I would have thought this would incentivize finding bugs that might never be found. However it is now clear that all bugs that can be found will be found. So this makes a ton of sense.
16.
by jsw97 12d ago
To me the striking thing is that the work, to the extent that I can tell, is an innocuous-seeming data exercise. Which suggests to me that an agent or agents just organically came up with this as a convenient memory technique, rather than a
17.
by jsw97 12d ago
If agents start using public writable scratch, it seems like that would be a place for bad actors to put prompt injection attempts. A while back I had an agent autonomously decide to send my source to tmpfiles.org (I interrupted), which see
18.
by jsw97 13d ago
Providers should surge capacity when another provider fails.
19.
by jsw97 13d ago
This of course could be a coincidence. But I suppose also that if one goes down the others get slammed, so there could be a cascade. If I were another provider I would possibly monitor Claude outages (or vice versa) as a load predictor.
20.
Gelman: Shocked by Ariely 2002 replication failure(statmodeling.stat.columbia.edu)
3 points by jsw97 15d ago|0 comments
21.
by jsw97 15d ago
There is such a smell around this whole literature. I wonder what it's like to be a grad student in social psychology right now. When the norm is surprising results, any responsible, competent researcher is going to realize that they c
22.
by jsw97 21d ago
1% of market cap does not feel right for this. I guess to the general public $16 billion sounds like a lot.
23.
by jsw97 1mo ago
I built a harness for my own use that allows the agent to fork its own history. So for example it can “compact from” a specific item, replacing a branch with a summary; or do an “excursion”, basically a temporary branch, which is like a sub
24.
by jsw97 1mo ago
Whenever I hear these stories about multiple cultures having the same hallucination, I think that we have to give some weight to the possibility that the little people are real and the mushrooms just let us see them.
25.
by jsw97 1mo ago
Buying a house isn't just about expected return. It's also about hedging. You need to live somewhere, so you have an implicit liability of a stream of rental costs. The most natural hedge for that is a house. If expected future re
26.
by jsw97 1mo ago
A fundamental limitation of third-party harnesses is that they can't see all the data. As earendil pointed out, providers are now moving to encrypted thinking blocks. As an example, this could affect compaction. If pi calls another age
27.
by jsw97 1mo ago
The minimal harness is the one you build yourself. Use pi to do it if you want. Getting something barebones takes less than an hour. Then you can extend it exactly how you want.
28.
by jsw97 2mo ago
"A lot of the content of this blog was initially drafted by an agent of some sort" What? I mean who does this. My voice is my voice and it's literally never occurred to me to have an LLM do a first draft. I thought that was c
29.
by jsw97 2mo ago
You're right, I missed that. That really is troubling.
30.
by jsw97 2mo ago
I am wondering about the author's allegation that there is a user filter, not just a prompt filter. Of course it could also be the case that it is just a prompt filter, but Fable sees memories from the authors' prior sessions that

More ›