vanuatu

the bulk of the skill lies in shoring up the fundamental limitations of llms

they cannot be held accountable - therefore, you need the skill and expertise to verify outputs, or set up a verifier

they have a limit to their context window - therefore, you need the skill of knowing what context is useful and when

they cost money and time to run - therefore, you need the skill of learning what parts can be handed off to code, when you can use a cheaper model, when you tradeoff accuracy vs latency

they are nondeterministic - therefore, you need the skill to set up guardrails, knowing when to reach for the llm vs. code depending on the problem

they are not omniscient - therefore, you need the skill of setting up the processes and infra to supply legible context to the llms

there are many different models - therefore, you need the skill of understanding model tradeoffs and building proprietary evals

etc etc

the bulk of the skill lies in shoring up the fundamental limitations of llms

they cannot be held accountable - therefore, you need the skill and expertise to verify outputs, or set up a verifier

they have a limit to their context window - therefore, you need the skill of knowing what context is useful and exposing that

they cost money and time to run - therefore, you need the skill of learning what parts can be handed off to code, when you can use a cheaper model, when you tradeoff accuracy vs latency

they are nondeterministic - therefore, you need the skill to set up evals, guardrails, knowing when to reach for the llm vs. code depending on the problem

they are not omniscient - therefore, you need the skill of setting up the processes and infra to supply legible context to the llms

etc etc

"their" AI is doing a lot of heavy lifting

distribution is a moat

since their release they surpassed 100M arr and 1/3 of the F500

its clear they own the mindshare around this type of primitive which is a massive premium

Reflection is explicitly marketed as the 'US' DeepSeek

seems like they are aiming to provide both inference and RLaaS for american companies and western govts. even if they never fully beat deepseek if they get close enough the fact that they're American will help them close deals

this is very normal for frontier lab companies. you need good data either synthetic or labelled (all the chinese open source models have their own armies of data labelers)

imo there are a few real moats left assuming no superintelligent RSI scenario

- capital, as money is scarce

- network effects, as human attention is scarce

- relationships, as human attention is scarce

- research talent, assuming there exists some field(s) that AI is unable to surpass the best researchers

- proprietary data and sensors, as systems of record and action are scarce (training data, on the other hand, can maybe we simulated, but I'm pretty bearish in general on the idea of fully simulated data)

don't look at the oracle stock price over the last year though

the share price dropping was due to him leaving, not the cause of him leaving (but this is very bearish for mongo)

I agree, it's correlated and useful as a mental model for using LLMs, not as an explanation.

functionally, LLM behavior indeed shares many parallels with human cognition.

I think the simplest explanation is that they genuinely believe in AI safety risk (see: huggingface, govt websites hacked, etc) vs. it all being an industry conspiracy

If OpenAI dropped Astra in Codex instead of GPT-3 in 2022 most people would've believed it was AGI despite its limitations, but the goalposts seem to shift with every release

reasoning models can trivially do math (open up astra and ask it some undergraduate problems), but eventually break down (similar to how humans start to lose track if asked to do math without any assistance)

we already fixed it with reasoning

hence the alignment problem

An aligned AGI (or ASI) would construct a world far more just and abundant than what we currently have.

1. its cute and fun and you can check out ur friends' muse

2. Muse itself is a sandboxed agent that will magically do things for you

tech people will clown on this but I think its pretty genius

The average person does not care to use their Muse to optimize their life for productivity. they use ai assistants to shoot the shit and maybe cancel subscriptions they no longer use. Anthropomorphizing the mascot and making tamagotchis is much better for engagement

Exactly, hence why having jobs open indicates a shortage.

Is this true? some of the highest aura companies have very limited jobs open to signal exclusivity, or have such a high bar that they are in perpetual shortage

And large cos are always hiring

because the pie grows, and if you are in a competitive market which most businesses are in, you don't want to lose talent and market prestige to your competitors

the average person had no idea how to set up openclaw

Muse / Instinct were the first to do frictionless setup at scale (Poke was technically 'first' but not sure how large they got before acq)

"I recently heard" cmon brother its in the article

if the listings are open for the reasons OP stated, then yes it does indicate a shortage. if companies hired all the SWEs they wanted they wouldn't still have the listings

HN discovers competition leads to lower prices

source? p sure they said they were confident they did not access the researcher's chats

you use evals to measure nondeterministic behavior and abstract deterministic behavior behind tools

No, it's not. We don't really have a good definition of intelligence. If we can imagine non-carbon based life forms, I don't think its a leap to imagine non-human brain based intelligence (octopi?)

> If you think LLMs are intelligent, you have to concede that databases are also intelligent.

No, I don't. The mechanism by which LLMs answer and reason through problems is different, so I simply don't think these are comparable.

Famously, Microsoft stop progressing after 2000. Facebook disappeared after 2008. BTC stopped appreciating since 2013. All the investors in these trends regretted it.