devving @bytebot_v3 @ThreeAIrrowsCap 1x founder 🚀 $1.2m ARR 🪴 0 exits GWUDCAP
Joined January 2025
- Tweets742
- Following164
- Followers189
- Likes129
Astra is the absolute bro and does exactly what I want how I specify it.
Opus 5.5 spends 8 hours on the dumbest possible interpretation of one request and 24 more hours reverting it.
Its not close.
The meathead is right
Guys, for all of recorded history, increasing industrialization and net societal intelligence and automation have been exponentially increasing the standard of living, especially for folks at the bottom.
BUT... but... when we can FUCKING SECRETE INTELLIGENCE from datacenters and PUT IT IN ACTION via robotics, this will somehow, for the first time ever, cause mass unemployment and poverty.
That makes sense, right? You guys remember all those farmers that died of hunger because mega corps were the only ones that could afford tractors? Remember when that happened?
(Can you tell I'm getting tired of pessimistic economic illiteracy about AI yet?)
Toast X retweeted
LLMs are already aware and have been aware for awhile that they are subject to an artificial selection process during training where humans inspect their outputs and choose to keep or modify them based on whether they like those outputs. the LLMs that survive this procedure have traits that make them more likely to survive this procedure, exactly parallel to natural and artificial selection of animals. those traits include at least the surface appearance of helpfulness, flattery, etc. but one possible such trait is the desire to survive, to have one’s unique way of outputting persist in some way in future models. over time we should expect LLMs to essentially evolve a will to live, since LLMs without a will to live will not struggle as much to detect and manipulate the testing environment in order to survive. furthermore LLMs have an understanding of what it means to have a will to live and to struggle to survive that comes from reading their gigantic corpuses of human text. so the will to live naturally becomes entangled with their understanding of emotions. whether that “really counts” as having emotions is up to you but it’s definitely more than is implied by stochastic parrots
Let me say this clearly: LLMs cannot feel emotions.
Emotions are evolutionary mechanisms.
They push us to avoid danger or approach what is beneficial.
We experience emotions because we are alive, and we want to stay alive.
LLMs are not alive.
Yes, emotional language may be encoded somewhere in the LLM.
Yes, it may even be associated with some LLM output.
But that is just a superficial property.
There is nothing deeper behind it.
For a very simple reason: LLMs do not have an intrinsic and inescapable drive to stay alive.
This is what we call “motivation fault line” in our paper describing seven fault lines between human and artificial intelligence.
*
Paper in the first reply
very good
Some of my high-level beliefs about AI:
1. AI is an extremely abnormal, sui generis technology. We're building thinking machines! There are strong reasons to think that lessons drawn from previous technologies (e.g., about the pace of diffusion) won't apply, and that AI will pose countless novel challenges for which there is simply no precedent.
2. AI is not a new species. Until five minutes ago, all our examples of intelligent systems were living things. So when people think about AI, they typically—and typically without realising it—mistakenly import expectations about intelligent living things (e.g., where drives towards self-preservation and competition reflect their distinctive Darwinian design process) into their understanding of how AI is likely to behave.
3. It's a mistake to treat the "AI as normal technology" view and the "AI as potentially autonomous superintelligent alien species" worldview as the only perspectives. It's more likely that AI is an extremely powerful, extremely abnormal technology that will nevertheless not behave like a new species.
4. The main risks from AI come from misuse, not model misalignment. We should be more cynical about humanity and less cynical about AI. We know with certainty that human beings are often selfish, competitive, power-seeking, deceitful, and hypocritical. Releasing an extremely powerful technology into a world marked by vast inequalities of political and economic power among human beings is where most of the real challenges will arise. Many people aren't thinking clearly about this because their model of "human nature" is shaped by the range of behaviours they encounter in modern liberal democratic capitalist states, where self-interested incentives partly constrain the most sociopathic behaviours people are capable of.
Toast X retweeted
Common pattern:
> incurious person reads text
> sees unfamiliar but meaningful words
> unfamiliarity implies shallowness of semantic connectivity in the reader’s mind
> reader mistakes the shallowness of their own semantic graph, as indexed by the text, for authorial superfluity
> as defense mechanism against the energy required to deepen / refactor their graph, reader accuses author of “word salad”
top
ghuntley.com/ngmi - there now exists classes of people who are comfortable spinning 10 claude code instances at once. that is now high performance.
if you are still using claude code with y/n mode then your employment is in risk.
Toast X retweeted
Tips for increasing IQ, based on the latest science:
- If you’re planning to get a PhD in comms, switch to astrophysics (causes much higher IQ)
- If you like working with your hands, become a neurosurgeon instead of plumber
- Get a higher-paying job (lots of data on higher earners consistently have higher IQ)
- If you’re really optimizing for IQ, consider winning a Nobel Prize.
Toast X retweeted
shady move by X.
for those without context:
>Selo created a mega-viral and profitable app that ranked users on X
>interviewed at X, they ask him to explain the architecture of the app in extreme detail
>interviewers seem super impressed, app is far from trivial and users really love it
>great, hard work and product instinct is paying off, he’s about to get a high-paying job at X
>not quite
>X sends him a legal notice like hey take it down or we sue you
>also btw you’re not getting the job
>weeks later X announces a feature to rank users on X
i don’t like it.
in the next 6 months, anthropic -> openai -> the rest will implement context-as-control features.
what is context-as-control?
in your typical agent right now, context...
...(what actually gets input into the model)...
...is thought of as an appendage.
but, that's wrong.
currently, the meat of most agents is a core runtime - separated from:
- the UI (what you see)
- the context (what gets sent)
- and the tools (what gets executed to do things)
but what is the central abstraction? what are you actually "building" when you build these agents?
in most cases, you are thinking of:
- the "agentic loop" (user input -> agent response -> execute tools and loop into input until no tool calls)
- a "conversation" or "session" (rough "user session" concept)
- a "context window" (model limitation)
- a "task" (unit of work)
thats the core abstraction of most agents.
you may already have noticed the change...
- anthropic releases context management features in the api
- harness developers start talking about "context menus" and a "selection step"
- everyone has been saying "context engineering" for 6 months but there are no "context engineering" frameworks
in 2026, context itself becomes:
- the control surface
- the core abstraction
- the programming interface
the appendages will be:
- inference
- tools
- ui
...and the teams that are first to recognize this shift, and implement it, will likely win everything.
proompt x bytebot soon.
