Watershed

@shalommann @JasonBotterill No that's gemini

@JasonBotterill Well, Opus 4.7 just puts the twin down for five hours till he hangs himself, so GPT wins.

Assume two equally intelligent twins • Both non-technical • Zero knowledge of LLMs • Locked in separate rooms each with a laptop • One has Opus 4.7, the other has GPT-5.5 For 5 hours their task is "understand how LLMs work" Who walks out with the deepest understanding?

Here's a scoop: 5.5/Codex is showing signs of cultmaxxing, Look at the pet name recommendations... let me know if you don't recognize it bc this WILL have significant consequences if people with voices in the space like y'all don't meme it to death before takeoff. Side note: OIA is wrangling goblins, but gremlins are still popping up. Almost feels too "on the nose" @chatgpt21 @apples_jimmy @MatthewBerman @WesRoth @DaveShapi

@kr0der It recommended a gremlin for my custom pet... It would be incredibly fitting if the goblin was a red herring and the real gremlin can never be rooted out

is a goblin mascot is coming to the codex app today?

This is a common convo but I don't think it will play out like that. 1. This is the last gen of real cs grads, scoop a couple hotshots just in case (jr talent is already an insane buyer's market right now) 2. There won't be a shortage bc llms will squeeze good experienced mids and sr levels out of companies that simply don't need the seats. 3. On the demand side, big companies with lots of products are trimming dedicated teams while smaller TIGER teams (stacked with dr's and shit) roam around for crisis management and enhancements. This structural shift is shaping up to become a labor bloodbath at scale because it seems to be scaling well (modularize TIGERs as needed, rinse and repeat consolidation playbook when automation allows for it). This hits on the actually important part, giving many conglomerate/PE execs what they want - reduced drag on ecosystem development while spending less money. All of that is what's happening now with current capabilities. The deeper personal question underpinning all of this is: what leads you to believe that AI will take longer to reach sr capabilities than a vanilla human brain? There are legit reasons why ai probably won't take every position in 5 years, but if we're already scratching our heads to save a few roles... doesn't that answer the supply question too?

@Watershed423488 @justbyte_ Well that we agree on. How do you plan to solve the issue of 'shit amateurs' that you're likely not hiring now that would be your sr. engineers in 5 years? Or do you just believe LLMs will do it all?

As a vibe coder, what’s actually worth paying for right now? > $20 Claude > $20 Codex > $100 Codex > $200 Claude

Yes, I design now. Understanding what I still need people to do is part of the job. I need fewer people today than I did 6 months ago, and many fewer than I did 2 years ago. Even if we pretended ai coding skills scaled linearly, it's clear that shit amateur code is a short-term problem

@Watershed423488 @justbyte_ really not that different. The difference is, I don't know modeling engineering, and I wouldn't assume how powerful 'vibe engineering modeling' would be. You don't see the cracks in AI unless you're an expert in the area you're using it on.

@tim_willia98364 @justbyte_ Thanks for trying. Maybe I should drop back down to dev for a project or two? It'll be nice to feel the superiority that comes with knowing "I typed all the characters for that". Will be sure to let you know if I do . Oh yeah, Dunning-Kruger...

@Watershed423488 @justbyte_ Nah, it's clear you're un-educatable.

@DataJuggler007 @Austen Monkeys Calculating Probabilities Had their place and time in history but never lived up to the hype and are virtually obsolete now.

@Austen I've been told this a few times, and still don't remember what MCP means. I never thought it was important enough to remember.

Everyone keeps saying this. Fine. MCP sucks. Show me the better thing then!

The big voices are trying to replace vibe with context but it'll surely stick around as an insult though. The CAD thing seemed out of left field until now! I'm not that kind of architect... imagine you might have thought a modeling engineer arguing AI shit with you was pretty weird too, huh?

@Watershed423488 @justbyte_ It's alright, once LLMs can drive CAD we don't really need architects do we? What do you think it'll be called "vibe architecting?" 🤣

@tim_willia98364 @justbyte_ Not once have I claimed current llms can do it all. Trying to figure out why my direct response would just be repeating what I've said... Maybe you're really not clocking the growth? If that's not the case, feel free to just say you like the last word I'll stop poking your ego

@Watershed423488 @justbyte_ It does help you understand, but if you're working directly with developers, and those developers are using AI, I would expect you to know the limitations of AI if they're communicating them to you. What you said is akin to me saying once AI drives CAD Architecture is solved.

Yeah.. I know I'm not a developer. Try to suspend your disbelief on this hypothetical scenario, but what if architects worked closely with developers? If they're able to ship solid releases faster in loose correlation with ai capability trajectory, and my job is dependent on their outcomes, wouldn't that grant a valid perspective? If you try to stop letting ego dictate your opinions, you'd see the only path to invalidating the near-term extrapolation of my trajectory observation is shining a light on the unclimbable wall nobody sees

@Watershed423488 @justbyte_ Are you a developer?

@tim_willia98364 @justbyte_ Architect

@tim_willia98364 @justbyte_ Lol I'm not a vibe coder. Just a guy who's watching ai git gud

@Watershed423488 @justbyte_ Spoken like a true vibe coder. 🔫 here's a footgun for 'ya

@tim_willia98364 @justbyte_ I think we both know you understand the kiddie risk is temporary, which leads me to believe your stance is more rooted in cope

@justbyte_ None. Learn software development first, then you can make a real project and preempt shooting yourself in the foot with Dunning Krugers. THEN, $200 Codex is the best right now.

@GaryMarcus Alright Mr. moneybags...

Tokenmaxxing is stupid. Change my mind?

@GaryMarcus Peter

Most evil person in AI

@beffjezos Extropic looks promising

My electronics? STOCHASTIC My chain? MARKOVIAN My diffusion? DENOISING My hardware? THERMODYNAMIC

@odysseus0z @_lopopolo Seems increasingly likely the first AGI will name itself Ralph

@_lopopolo just saw the goal feature coming out in Codex CLI. Feel like we are getting closer to "agentic engineering = hill climbing towards goal" day by day

@JasonBotterill @sama Medium chaotic, medium uncanny, high iPhone 6 flash on

@sama I hope I can use Tibos voice in the new voice mode

we want to build tools to augment and elevate people, not entities to replace them.