@femalenatured @RealGame101 these are 10000% fake text. lmao. anyone who has ever texted a girl or messaged on dating app know these are 10000000% fake.
English
Chris
2.6K posts
Chris
@Christiano92
A powerful Unreal Engine 5 plugin that implement gameplay features for you across Blueprints, Anim Blueprints, Widgets, Inputs, Materials | Universal Game Pilot
Katılım Mayıs 2009
474 Takip Edilen1.1K Takipçiler
Logistics only with @RealGame101’s signature “Invite me over” for when you’re out-of-town.
Numbers Game.
English
@NorthDaygame Guys anyone telling u to go on coffee (day) dates is lying to you so they can sell u stuff. Theres nothing more platonic than a coffee day date. Don’t get fooled by fake AI text pics. Nothing beats nearby cheap restaurants/drinks > escalate during walk/car > crib.
English
how my first dates looked when i was a virgin:
-2 hour dinner at an expensive restaurant
-i paid for everything
-i picked her up
-i ran out of things to say at minute 40
-i texted "i had a great time :)" the same night
-she replied 2 days later "you're sweet"
-i never saw her again
how my first dates look after 411 dates:
-30 minute coffee 4 minutes from my apartment
-i pay for everything
-i tell her where to meet
-i sit across from her, not next to her
-i check my phone in front of her
-i say my "pull" line when the coffee ends
-we go for a walk
-the walk ends at my place
-she ends in my bed
is this guaranteed? fuck no.
we still fuck it up some times.
what it is guaranteed is that:
--> before i didn't pull and bang ANY girl
--> now I pull and bang many of them
focus on improving.
English
Block me. who cares. this is why alot of people have issues with you. its not what you say its the pompous way you are saying. always "i'm right", "give me credit", "you're wrong", "You're XYZ". Basically you are saying "i'm the arbiter of truth if you don't agree with me im blocking you and saying you slandered me". You leave no gray area, no nuance, no reasoning. You say OpenAI made it clear they are for mass surveillance. If they sued you for defamation. They would argue the complete opposite, that in-fact they have been clear that they are AGAINST mass surveillance. Perfect example i tried to go in-depth with you regarding Waymo & SDC and the architecture used and instead of engaging on the actual detailed arch you completely dismissed it outright claiming you need a personal or w/e statement directly from Waymo and then pointing to an experiment that was completely irrelevant to the discussion. The fact that you don't see that OpenAI will vehemently argue/contest against "openai will sell you private data to the government" and attack someone who states that they would, shows just how much you feel like you are the arbiter of truth.
English
@Christiano92 @tszzl what is inaccurate about what i said? they have made clear they are open to mass surveillance and they have lots of private data about mental health, physical health, etc.
that’s factual; your tweet is what is defamatory. will block if you don’t retract or better explain.
English
nope you are.
because openai will sell your most private data to the government.
Greg Brockman@gdb
the model alone is no longer the product
English
@GaryMarcus I'm not talking about EMMA which was just a experimentation just like ChaufferNet, its not at all part of their system in any way. I'm talking about the Waymo Foundation Model. But I get it, no human on earth will change your mind Not even Waymo. waymo.com/blog/2025/12/d…
English
1. read the list of issues discussed here: waymo.com/research/emma/
and
2. Know that the New York Times fact checked my understanding of this with *Waymo* in October 2025 and Waymo said they had no issues.
I am not going to believe an anon without strong sources over that, sorry.
And yes I do agree that ML has been in used in many aspects throughout. But as of October 2025 Waymo’s worked mostly fine in a modular neurosymbolic fashion with LLMs etc only as components; not end-to-end. Waymo confirmed this.
Good day to you, unless you can *source* your statements.
English
it’s funny how people here just make stuff up.
Chris@Christiano92
@GaryMarcus Take away LLM & every single AI application today goes back to the stone age, driverless cars will immediately break down, all the apps will stop working. it would be impossible to get them back up. It simply will not work. Progress is clearly due to LLM not irrelevant of it.
English
LLM tech is used in different aspect of the self driving stack, from training to generating datasets, simulation, validation models, inference driving models. Lets take the prediction stack arch for example. How has it evolved over the years? Do you even know? It went from Grid CNN behavior prediction > MultiPath >MultiPath++> StopNet > SceneTransformer > JointFuturePrediction > Wayformer > MotionLM
English
two things i can’t find in the only very short report (a waymo blog) I could find with that figure: 1. have the symbolic components been replaced (an earlier paper pointed to many challenges in using the pure VLM apprach) and 2. does the VLM outperform the alternative (over a wide range of circumstances). the blog doesn’t address either.
English
@pythagoras999 @GaryMarcus From training, validation, inference, LLM tech has invaded every part of self driving. Even the models that make up a typical self driving system (perception, prediction, planning) now use LLM like architectures.
English
@pythagoras999 @GaryMarcus You need to educate yourself. Waymo literally uses Gemini
waymo.com/blog/2025/12/d…
English
I love AI, it’s pure LLMs I hate.
Pure LLMs *are* basically just autocomplete.
Recent progress (e.g. Claude Code) doesn’t show otherwise
Rather, lot of the progress in the last two years has come from *introducing* other things – mainly classic symbolic techniques and tools, to offset the weaknesses of pure LLMs.
Shame to see the tweet below muddle all of this.
If we want to make further progress we need to understand where the progress is coming from; mostly it is coming from leaving pure LLMs behind.
Bindu Reddy@bindureddy
Where did all the AI haters go? 🤔 You know, the ones screaming "it's just autocomplete!" and "it'll never be useful!" They're real quiet now that AI is actually magical and transforming everything. Almost like... they were wrong the whole time. 😏
English
@GaryMarcus Its not experimental its in the stack, on the car & huge part of training & validation. They literally leverage Gemini & they clearly state it & that their biggest boast in performance has been from LLM in their tech talks. You just choose to dismiss it. waymo.com/blog/2025/12/d…
English
you need both for optimal function; that’s the claim i have been making for 30 years. neuro+symbolic.
as a point of fact though, most of waymo’s LLM stuff at least as of last summer was still experimental. i am not actually sure how much lift they are getting from LLMs in production, though they are actively trying them. but they had them running for years without any LLM in production.
i do know that the New York Times fact checked my description of their tech in my October oped that mentioned Waymo and that Waymo had no objections.
English
@GaryMarcus lol you didn't win nothing. you clearly lost and just continue to shift the goal post. For example the biggest breakthrough in self driving (WAYMO) was utilizing LLM and foundational models. Its what's lead to the huge increase in performance, capability and deployment scaling.
English
The pure LLM debate - which I had for many years, here and elsewhere - is indeed no longer relevant. Why?
Because I won; nobody uses pure LLMs anymore.
Nowadays all deployed objects are neurosymbolic, which was exactly the point of my infamous 2022 paper, Deep Learning is Hitting a Wall.
If you don’t know I won, it’s because you read the title and not the paper 🤷♂️
MachineSovereign@VizierPrime
This is true in the narrow sense, but it also points to why the “pure LLM” debate is becoming less central. The deployed object is not a pure LLM. It is a language model embedded in a tool-using execution stack: retrieval, code, memory, verifiers, APIs, agents, symbolic constraints, workflow permissions, and external systems. So the question is not whether autocomplete alone becomes intelligence. The question is what happens when autocomplete becomes the interface layer for systems that can act, check, search, write code, call tools, route tasks, and operate inside institutional workflows. That hybrid object is what matters. Aviation was not transformed by “pure engines” either. It was engines plus control surfaces, navigation, fuel systems, airports, maintenance regimes, regulation, and logistics. AI will likely be the same. The model is not the civilization-level unit. The stack is.
English
@thsottiaux @doublemover You need to fix this, mine actually reduced usage, it went from ~45% to 30%. so there has to be a bug or something.
English
I logged on to say thanks but mine isn't reset
Tibo@thsottiaux
Codex usage limits have now been reset across all paid plans. Enjoy the weekend!
English
@terjeholden @spleck Gotta love the Tesla fans. ~1-2 cars in each city for PR is OMG THEY ARE EXPANDING FAST. 10 TRILLION STOCK. BUY BUY BUY! From 1 million to 1-2 cars, from Entire country to a few blocks. I'm so sorry im in denial i will get on my knees and apologize! reddit.com/r/SelfDrivingC…
English
@Christiano92 @spleck Holding on? Tesla launching robotaxis in 3 cities unsupervised so far, and expanding. Are you in denial?
English
Why would you start an unsupervised region without validating the maps first? These are *small* geofences. Doesn’t seem unreasonable to give them a once over. That’s multiple map based failures in the first few hours of operation. :(
TexasTSLA@TexasTSLA
Guess @robotaxi in Dallas needs a map update like the rest of us FSD users 😅😅 It tried to drop me off 2.6 miles away from my destination and got stuck in an infinite loop but thankfully rider support took over 🕹️😀 ✌️❤️
English
@terjeholden @spleck lol after 11 years you are still holding on for dear life. whatever they put in that koolaid must be STRONG.
English
@spleck I mean going in a loop is not a safety issue, its a convenience issue as you just let rider support know it cant find a way out. Also the highway glitch sort of shows how good the system is, it didnt freeze or stop on the highway, no panic, just looking for nearest exit
English
LLM agents need a working memory that can be injected into their thinking token without forcefully steering/interrupting them. Currently they don't follow skills or MD to the T and sometimes completely ignore it. The first LLM to engineer this gets 100x useful @thsottiaux
English
@thsottiaux No cause it will be penalizing ppl for working during the day. It just turns into Uber. Peak pricing never works out for consumers
English
With Codex the there is quite the gulf in load between peak and off-peak times, and we would like to achieve more of a smoother traffic pattern as that would be a more optimal use of our compute.
We have ideas, but curious what you all think we should do? Would more usage during off-peak and surge multiplier during peak times make sense?
English
@rohanvarma how about an unlimited slow mode for $200 PRO users when we run out of tokens?
English
If we made /slow mode in Codex, would you use it? What for?
(Slower inference at a cheaper cost)
English