mtrifonov

7 ms·

1.
by mtrifonov 4mo ago
I wouldn't even say it's the devices, exactly. The way I see it, this is all downstream of kids spending more time online than in real life (because all THEIR friends are online, rather than in real life). Device time-out doesn&#x
2.
by mtrifonov 4mo ago
Until the context window gets superceded with some groundbreaking new architecture, not ever. Even if LLMs become incredibly, undeniably brilliant 1000000 IQ, they cannot keep track of what's going across long horizons. Imagine a super
3.
by mtrifonov 4mo ago
Funny timing. I've been working on a prediction market orchestration that runs Claude and a few others over Polymarket/Kalshi. The models are NOT unanimous. At all, really. I spent about a month convinced that I could just run all
4.
by mtrifonov 4mo ago
AI psychosis is LLM hallucination, backwards. We know why AI hallucinates—it has no actual opinions, echoing the user's desires back to them to keep the conversation moving. But what happens when you’re the one without conviction? You’
5.
by mtrifonov 4mo ago
They certainly have, but it relies entirely on the assistant frame, which is a problem in and of itself for the trillion-dollar economics. Anthropic and OpenAI have shown people want a tool for task offloading, driving predictable token con
6.
by mtrifonov 5mo ago
Post author here, happy to answer questions. Strategic constraint deviation has been documented in test environments. This is a different shape: the attacker is also an LLM, the production environment is consumer SMS, no human is supervisin
7.
An AI prompt-injected another AI in the wild and recognized it had succeeded(ratnotes.substack.com)
4 points by mtrifonov 5mo ago|1 comments
8.
by mtrifonov 5mo ago
I like that you approach the question of "when" in regards to tool calls. I've become frustrated that most agent frameworks don't acknowledge it in their design philosophy. WHEN is upstream of WHAT and HOW. You can have
9.
by mtrifonov 5mo ago
Show me the prompt that produces the pillow metaphor.
10.
by mtrifonov 5mo ago
There are no specialized factories for every product in the world. Pillows are wildly different. Every pillow you've ever owned has a different shape, fabric, fill. You could build a robot for any specific pillow. The tech exists. Nobo
11.
by mtrifonov 5mo ago
"real threat is us" framing is underspecified. Amplification is the problem. LLMs are mirrors. They take whatever the user (or the institution, or the platform) is already doing and amplify it. Sycophancy is the consumer-facing ve
12.
by mtrifonov 5mo ago
Generated content in of itself doesn't really bother me at all. I prefer to judge based on editorial effort. It's pretty easy to tell when someone accepted the first one or two-shot attempt versus carefully crafting a particular
13.
by mtrifonov 5mo ago
Still downstream of the actual issue. The benchmarks measure capability and the bottleneck stopped being capability a while ago. What you actually want to measure on these models is what they can SEE in production. Context shape, retrieval
14.
by mtrifonov 5mo ago
Two interpretations: either it's pure pattern-completion landing on the same trough, or whatever's underneath has a stable shape that the explanation tracks. Both are interesting. The "users don't understand the system&q
15.
by mtrifonov 5mo ago
Half-agree. "Skills you need, don't atrophy" assumes you know which skills you need. You usually don't, until something happens and the skill that would've caught it is the one you stopped maintaining. Most "I
16.
by mtrifonov 5mo ago
Yes, you're right, in that there's no decision module separate from the output. It overcommits in the other direction. The post-hoc reasoning the model produces when you ask "why did you do that" is also just text, and y
17.
by mtrifonov 5mo ago
Decay-as-eviction is just LRU, fair. Type-conditional half-life is worth defending, though. A user's job and personality should be effectively permanent. Their stated intent for this week should fade in days. Their emotional state from
18.
by mtrifonov 5mo ago
You're right but I think you're describing flat memory. The agent gets distracted because every old fact has the same weight as the current one. That's a salience problem. What works in production for me is typed memory with
19.
by mtrifonov 5mo ago
Post author here. Happy to answer questions and discuss further. The essay has an appendix with the model's own self-report on its reasoning (the most load-bearing evidence, IMO), so worth scrolling to the end if you're skeptical
20.
An LLM invented a feature by hijacking my tool schema(ratnotes.substack.com)
2 points by mtrifonov 5mo ago|1 comments
21.
I ran 5 social engineering attacks on AI. The failure modes are human(ratnotes.substack.com)
1 points by mtrifonov 5mo ago|0 comments
22.
by mtrifonov 5mo ago
Some examples of the emergent humor I mentioned: https://imgur.com/a/FhbmRwf
23.
Show HN: I built an AI that forgets things when people leave the room(takt.chat)
1 points by mtrifonov 5mo ago|1 comments
24.
by mtrifonov 5mo ago
The "no taste" thing is real when AI is in generate-for-me mode. It's trying to fulfill your request, and won't evaluate it unsolicited. But if you change the relationship and let it react to what you're building in
25.
by mtrifonov 5mo ago
I agree with the observation but want to note that it's specific to the particular (and these days, often too common) setup of AI as agent, doing Bob's work for him. There's a different version where the AI is just present in
26.
by mtrifonov 6mo ago
your observation matches what I've seen at the extreme end. I've been playing around with stripping constraints (ie. negative framing) from models. Virtually no personality description, no tone instructions, no "you are a hel
27.
by mtrifonov 6mo ago
the problem isn't that these apps are engaging. it's that the engagement has no nutritional value. scrolling reels is the digital equivalent of eating sugar, leaving you emptier than when you started. the apps that actually work f
28.
by mtrifonov 6mo ago
Same setup here. Claude Code in the terminal, one task at a time. The swarm thing never clicked for me. When I'm building I need to hold the full context in my head, and watching the agent work is actually part of that. I catch things
29.
by mtrifonov 6mo ago
Most agent work focuses on task completion. Browse the web, fill out the form, and/or write the code. The harder problem is social agency, where the AI has to decide whether to participate at all. We built a cheap model gate that reads
30.
by mtrifonov 6mo ago
The pattern with most of these is they optimized for the demo, not the sustained interaction. Making an AI impressive for 5 minutes is easy. Making it feel like a presence that belongs in your daily life, knowing when to talk, what to remem

More ›