Fable 5.2 had to be crippled...
Credibility score: 49/100 β Mixed Credibility. Several questionable claims detected. Watch with healthy skepticism.
BSmeter analyzed "Fable 5.2 had to be crippled..." and rated it 49/100 for credibility (a BS score of 51/100 β mixed credibility), on 2026-11-10. Its weakest claim β "Will other labs follow suit or 'release all it out there?' β a false dilemma, as if those are the only two options. π" β scored 20/100 and was flagged as false dilemma. 35 claims were checked against the video transcript. Scores are produced by BSmeter's AI analysis of the transcript, not independent human verification.
Of 35 claims analyzed: 2 scored under 40, 27 between 40 and 69, and 6 at 70 or above.
Claims analyzed
Declares it the 'best model in the world' with zero evidence. Bold. β Confidence Mismatch (45/100)
At 0:04
He says 'best in the world' like it's a fact, but offers no comparison or data. That's not proof β that's just a strong opinion. π
Why this score: Calling something 'the best in the world' without any benchmarks, comparative data, or even a specific metric is pure hyperbole. It's a statement of extreme confidence that isn't backed by any presented evidence, making it an assertion of personal belief rather than an objective claim.
Original quote: βNow, without any exaggeration, this is the best model in the world, at least from models that are available to us mortals.β
Sources: Claude Sonnet 4.5 is probably the βbest coding model in the worldβ (at least for now), Anthropic Claims 'Best Coding Model in the World' With Claude Sonnet 4.5βWe Tested It, Bold hypothesis - Wikipedia
Claims it 'speedruns' tasks Opus 5.5 can't, no specifics. Just vibes. β Confidence Mismatch (45/100)
At 0:12
He says it 'speedruns' what Opus 'stalls' on, but doesn't name a single task. That's not a comparison β that's just a flex. π
Why this score: The speaker makes a strong, specific claim about performance superiority over a named competitor (Opus 5.5) without providing any concrete examples of tasks, metrics, or scenarios where this 'speedrunning' occurs. It's a confident assertion of performance without any verifiable details to back it up, leaving the audience to take his word for it.
Original quote: βIt speedruns things that humans are incapable of doing, even that other very advanced models like Opus 5.5 are just not capable of doing. Like, Opus stalls and this thing just speedruns it.β
Claude's results were checked by a physicist, then by another Claude model. That's not how verification works. π β Confidence Mismatch (45/100)
At 0:42
They say a physicist 'checked it,' but the real 'verification' came from another AI. That's an echo chamber, not proof. π
Why this score: The speaker implies human oversight, but then immediately undercuts it by stating the primary confirmation came from 'another Claude computation.' It's like asking a child to grade their own homework, then having their twin brother confirm it. That's not independent verification; it's a confidence trick.
Original quote: βNotice an internal claude research model obtained the results presented here. The physicist, you know, checked it, tried understanding it, and then notice what they say next. Another claude computation using a different method not included here gives the same result.β
The speaker's personal 'problem' will be solved by Fable 5.2 before the video ends. That's a bold claim. π₯ β Confidence Mismatch (45/100)
At 0:56
He's hyping up Fable 5.2 by claiming it'll solve his personal problem mid-video. That's not a demo; it's a sales pitch with a timer. π
Why this score: The speaker uses his personal experience with a 'problem' he's been trying to solve as a dramatic setup for Fable 5.2's capabilities. Claiming it will solve it 'before I finish recording this video' is a classic hype tactic, designed to create a sense of immediate, almost magical, efficacy without actually demonstrating it. It's a promise, not proof.
Original quote: βSo Claude came up with it. A human looked at it and tried to understand it. But we know it's okay because another Claude model said, "Yeah, the math checks out." So everyone's losing their mind as to how powerful this model is. Notice this says contributed by Anthropic. My results have beenβ¦β
Anthropic 'specifically crippled' Fable 5.2 during training. That's a strong accusation with zero evidence. π β Loaded Language (45/100)
At 1:56
He says Anthropic 'crippled' the model. That's not an observation; it's an accusation of malicious intent, completely unbacked. π
Why this score: The term 'crippled' is highly charged and implies deliberate sabotage or severe limitation. Without any evidence or explanation of *why* Anthropic would do this, or *how* it was done, it functions as pure loaded language designed to evoke outrage or suspicion. It's a dramatic claim presented as fact.
Original quote: βThis model is very different from what we've received in the past and I think all the Frontier models released after this will be similar because with all of the previous models, the AI labs tried to release the best model they could. It was a race. How do how do they turn their billions of dollarsβ¦β
Claiming Anthropic 'crippled' models to hide capabilities β pure speculation, no proof. π β Confidence Mismatch (45/100)
At 2:30
He says 'crippled' and 'didn't want out there' like it's a known fact β but it's just a guess. π₯
Why this score: The speaker presents the idea that Anthropic intentionally 'crippled' their models and 'didn't want' certain things revealed as if it's a confirmed motive, when it's pure conjecture. There's no evidence provided to back this up, just a confident assertion of intent.
Original quote: βcrippled certain parts of this model during training so they wouldn't reach their full capability. And also, we might now know what it was that Anthropic didn't want out there.β
Citing the 'Meter report' as direct evidence β but only showing a logo, not the actual data. π β Missing Context (45/100)
At 2:37
He points to a logo and says 'directly from there' β but where's the actual report, mortal? Show the receipts. π©
Why this score: The speaker references a 'Meter report' and claims it's 'directly from there,' even showing a logo. However, he doesn't display the specific data, charts, or text from the report that supports his 'crippled' claim. It's an appeal to an external authority without presenting the actual evidence.
Original quote: βHere's the thing. This model is model 3 early. As you know, a few days ago, the meter report came out. This was directly from there. Here, there's their there's their logo, meter. I'm not hiding it.β
Dismissing Anthropic's explanation with 'probably not the case' β no new evidence, just vibes. π β Confidence Mismatch (45/100)
At 2:56
He says 'probably not the case' to dismiss Anthropic's own statement. That's not a rebuttal, that's a shrug. π₯
Why this score: The speaker dismisses Anthropic's original explanation for a capability jump ('maybe it was just a one-time jump') with a confident 'that's probably not the case.' He offers no new evidence or data to contradict Anthropic's statement, relying instead on his own conviction.
Original quote: βSo, some people were concerned that right around the Mythos release, there was a rather sudden jump in capability. Originally, Enthropic said maybe it was just a one-time jump and then it kind of continued in the same way. But, as you can see now, that's probably not the case.β
Using a 'high school test for college student' analogy β a relatable but unproven comparison. π β Loaded Language (45/100)
At 3:11
The analogy is cute, but it's a feeling, not a fact about how these models are actually measured. π©
Why this score: The speaker uses a vivid analogy of giving a 'high school level test' to a 'college student' to explain the difficulty in measuring Model 4's progress. While relatable, this is an emotional appeal and a simplification, not a precise technical explanation of benchmark limitations or how Anthropic actually measures its models. It's designed to make the model sound more impressive without concrete data.
Original quote: βSo, in one of the previous videos where we covered this meter report, we focused a lot on model 4. Model 4 is unique. It's crushing anthropics internal benchmarks. It's getting a little bit harder to measure its progress. If you can only give somebody a high school level test, but they're alreadyβ¦β
Claiming 'all results were above 50%' on an internal benchmark β without showing the benchmark or results. π β Anonymous Authority (45/100)
At 3:34
He says 'all results were above 50%' on a benchmark he 'forgets the name of.' That's not data, that's a memory lapse. π₯
Why this score: The speaker claims 'all of the results were above 50%' on Anthropic's internal AECI benchmark, but doesn't show the benchmark itself, the specific results, or any verifiable source for this claim. He even admits to forgetting the name initially, which further weakens the authority of the statement. It's a vague reference to internal data without actually presenting it.
Original quote: βModel 4 was the first model to really cross into that territory. All of the results were above 50% on I forget the name of the benchmark the internal benchmark at Entropic AECI that's what it is. Apparently Enthropic had beyond what was available to us model 2 model 3 and model 4 that model 3 andβ¦β
Claiming the gap between current models and the 'frontier' is 'razor thin' β a confident, unproven assertion. π β Confidence Mismatch (45/100)
At 4:00
He says the gap is 'razor thin' like he's seen the internal roadmap. That's not insight, that's a hunch. π©
Why this score: The speaker confidently asserts that the distance between publicly available models (Fable 5.2/Mythos 5.2) and the 'frontier' (Anthropic's latest internal models) is 'razor thin.' This is a subjective assessment presented as fact, without any concrete data or evidence to quantify this 'razor thin' gap or how he would know its exact width.
Original quote: βremember fable and mythos they're pretty much the same. Fable just has more guard rails in place. So, if these are the Model 3 check marks here, this is the latest and greatest internal super secret benchmark breaking one. We're getting the thing like right there. I was wondering if the distanceβ¦β
Anthropic found a secret to speeding up AI, but they won't name it. π β Anonymous Authority (45/100)
At 4:30
They're hinting at a massive secret without naming it or providing any specifics. It's the oldest trick in the book. π
Why this score: The speaker is building suspense around an unnamed, powerful discovery by Anthropic. This creates an air of mystery and importance without offering any verifiable details about what this 'thing' actually is, making it impossible to scrutinize. It's a classic 'we know something you don't' move.
Original quote: βrelease was wondering what is it that Anthropic discovered that seems to be really speeding up AI development because we know that's happening. and they've said it, but they were extremely careful not to hint at the actual name, the thing that's causing that acceleration.β
Anthropic redacted a report to hide the AI speed-up, nearing 'recursive self-improvement.' π β Confidence Mismatch (45/100)
At 4:42
They're claiming Anthropic is 'nearing RSI' and 'maybe past that threshold' with zero actual evidence. That's not a fact, it's a wish. π©
Why this score: The speaker attributes a strong claim about Anthropic's progress towards 'recursive self-improvement' (RSI) and potentially 'past that threshold' to Anthropic itself, but without providing any direct quotes or specific reports. The use of 'maybe' and 'at least approaching it' softens the claim, but the overall implication is presented with unearned certainty about a highly speculative concept.
Original quote: βSo, here's Sam Mcandlish, one of the co-founders of Enthropic, saying they redacted a lot from the report because they don't want to share exactly what's driving this speed up. So, Anthropic is saying that they're approaching RSI, recursive self-improvement, and that model for something rightβ¦β
Anthropic is limiting its models' gains, despite them being 'smarter and stronger.' π β Volume Game (45/100)
At 5:03
First, it's 'blowing people away,' then it's 'limiting gains.' Which is it? It's a classic bait-and-switch. π
Why this score: The speaker first hypes up the models as 'generally smarter and stronger' and 'blowing people away,' only to immediately pivot and claim Anthropic is 'beginning to actually limit the gains.' This creates a contradiction where the initial positive framing is undermined by the subsequent negative one, leaving the audience confused about the true state of the models' capabilities. It's a rhetorical trick to keep you hooked.
Original quote: βSo, how's it possible that they're releasing this something that's that much more generally smarter and stronger and it's just blowing people away? Well, specifically, Anthropic is beginning to actually limit the gains that these models can make.β
Fable 5.2 and Mythos 5.2 are at Opus 5.5's level, which is 'the peak' for certain areas. π β Confidence Mismatch (45/100)
At 5:19
Declaring Opus 5.5 'the height' and 'the peak' of abilities in 'certain very specific areas' is a bold claim with no backing. That's not a fact, it's a prophecy. π©
Why this score: The speaker makes a definitive statement that Opus 5.5 represents 'the height' and 'the peak' of model abilities in 'certain very specific areas.' This is presented as an absolute truth without any supporting data, benchmarks, or expert consensus. Such a strong declaration about a rapidly evolving field like AI is highly speculative and lacks the necessary evidence to be presented as fact.
Original quote: βFable 5.2 and Mythos 5.2 perform about the level of Opus 5.5. Not for everything obviously, or only certain very very specific areas. And that's important to understand because we're probably going to see this moving forward. Opus 5.5 was the height of models abilities in these particular areasβ¦β
The 'gimping' isn't user-facing suppression, but deeper architectural limits. π β No Frame (75/100)
At 5:45
He's clarifying the technical distinction between runtime suppression and foundational architectural limits. It's a clear explanation. π₯
Why this score: The speaker is providing a clear and specific distinction between two different ways AI models could be limited: at the inference (user interaction) level versus at the foundational architectural level during training. This clarification helps the audience understand the technical nuance of the previous claim without employing any manipulative framing.
Original quote: βWhat are those areas? Well, it's things like accelerator kernel optimization and improving large language model architectures. So this isn't them, you know, altering the capabilities as the model is being served at inference time, right? So, as the users are typing in questions, something's gettingβ¦β
Anthropic limited 'specific research capability' during training, but the model is still a 'beast.' π β Volume Game (45/100)
At 6:02
First, they 'limited advances' in 'specific research capability,' then it's an 'absolute beast' that 'most of us won't run into issues' with. Pick a lane, mortal. π
Why this score: The speaker quotes Anthropic admitting to 'limit advances in specific research capability' during training, which implies a significant restriction. However, immediately after, he downplays this by saying the model is an 'absolute beast' and 'most of us won't run into any issues.' This creates a contradictory message, where a potentially concerning limitation is quickly dismissed, making it hard for the audience to gauge the true impact.
Original quote: βThey're saying, "During the model's training, we took steps to limit advances in specific research capability that could most directly accelerate AI development." Now, don't get me wrong, this thing is an absolute beast. Most of us won't run into any issues with what they've restricted.β
Fable 5.2 was 'purposefully gimped' in development to prevent Claude from having certain abilities. π β Loaded Language (45/100)
At 6:18
Calling it 'purposefully gimped' is pure emotional language, not a neutral description of development choices. That's not analysis, it's an accusation. π₯
Why this score: The speaker uses the highly charged term 'purposefully gimped' to describe Anthropic's development choices. This loaded language implies malicious intent or deliberate sabotage, rather than a strategic decision or a technical limitation. It's designed to evoke a strong negative emotional response from the audience, framing the situation in a highly critical and accusatory light without objective evidence of 'gimping' in the pejorative sense.
Original quote: βNot even restricted, but purposefully gimped in development so that Claude doesn't have these abilities once it comes out. For example, as an experiment, one anthropic engineer gave a Fable 5.2 the task of designing aβ
Claude built an OS from scratch in 8 days β a bold claim of AI capability. β No Frame (75/100)
At 6:30
Claiming an AI built a full OS in 8 days from an empty folder is a hell of a flex. If true, it's impressive.
Why this score: The speaker presents this as a direct achievement of the Claude model. While the specifics of 'operating system' and 'working' can be vague, the core claim of rapid, autonomous OS development by AI is a significant one. It's presented as a factual report, not a hypothetical.
Original quote: βa Fable 5.2 the task of designing a prototype operating system. So, starting from an empty folder, right? So, Claude is handed an empty folder. The model coordinates over 1,000 agents working around the clock for 8 days writing, testing, and reviewing code. It produced a working operating systemβ¦β
Mentions 'some people complaining' about Claude's reasoning, then immediately pivots to Anthropic's 'anti-distillation' efforts. β Volume Game (45/100)
At 6:58
He throws out a complaint about 'stupid mistakes' then immediately buries it under a 'can't verify' and a pivot to 'anti-distillation.' Classic misdirection. π
Why this score: The speaker introduces a negative point ('stupid mistakes,' 'fake tools') attributed to 'some people' but quickly dismisses it with 'I personally haven't encountered that, so I can't verify.' This minimizes the complaint's impact. He then immediately shifts to Anthropic's efforts to prevent 'distillation,' implying a different, more complex reason for any perceived issues, rather than the model's inherent flaws. It's a quick pivot to control the narrative.
Original quote: βApparently, the sort of reasoning traces, the chain of thoughts that these models are putting out. Some people have been complaining that through those reasoning traces, clot thinking through stuff, they don't make any sense. They they make stupid mistakes they call fake tools. I personally haven'tβ¦β
Claims open-source models 'feel surprisingly like' top models, then offers a 'wild stab' theory about autonomous AI swarms. β Confidence Mismatch (45/100)
At 7:27
He says 'wild stab in the dark' but then lays out a whole detailed theory about AI swarms and mega-structures. That's not a stab, that's a manifesto. π
Why this score: The speaker prefaces a significant speculative theory with 'I'm going to take a wild stab in the dark' and 'Please feel free to correct me.' This lowers the audience's guard, making the subsequent, highly speculative claims about 'autonomous swarms of AI agents running across compute cluster mega structures' seem less like definitive statements and more like casual speculation. However, the detail and conviction with which he describes this future vision contradict the initial humble framing, creating a confidence mismatch.
Original quote: βThat's why we see sometimes open source models coming out that that perform and act and feel surprisingly like the current number one model on the leaderboards. So here's what I think is happening. I'm going to take a wild stab in the dark. Please feel free to correct me in the comments, but theβ¦β
One researcher's opinion is presented as a definitive endorsement of rapid AI progress. π β Anonymous Authority (45/100)
At 8:34
He cites 'one research' endorsing 'greater than two years of progress in one year' and then says 'four is not crazy.' β That's not data, that's a single opinion amplified. π
Why this score: The speaker uses a vague reference to 'one research' (singular, not plural, which is odd phrasing) to support a significant claim about AI acceleration. This 'research' is unnamed, uncited, and its methodology or context is completely absent. Then he extrapolates from 'greater than two years' to 'four is not crazy' without any further evidence, presenting a personal speculation as a reasonable extension of an unverified claim.
Original quote: βOne research said they would definitely endorse a greater than two years of progress in one year. And I don't think four is crazy.β
He claims releasing advanced AI is 'handing out a self-replicating machine' β a dramatic leap. π© β Loaded Language (45/100)
At 9:51
From 'autonomous AI research' to 'self-replicating machine' in one breath. β That's not a logical step, that's a horror movie pitch. π
Why this score: The speaker uses highly charged, fear-inducing language ('self-replicating machine') to describe the release of advanced AI capable of 'autonomous AI research.' While the capabilities of AI are a valid topic for discussion, framing it in such an extreme, almost sci-fi, manner without concrete evidence or a clear explanation of how 'autonomous research' directly translates to 'self-replication' is designed to evoke an emotional response rather than provide a balanced assessment. It's an appeal to fear, not logic.
Original quote: βis somewhere here, let's say there seems to be a line. If an AI lab releases something beyond this capability in certain very specific domains, they're basically handing out something that can do that sort of autonomous AI research. So they're handing out a selfreplicating machine.β
Anthropic is drawing a line, won't push models past Opus 5.5 performance. That's a bold prediction with zero receipts. π β Confidence Mismatch (45/100)
At 10:32
He's stating Anthropic's future strategy as fact β but it's pure speculation, no actual announcement. π₯
Why this score: The speaker claims Anthropic 'will no longer push these models' past a certain performance level, presenting it as a decided strategy. This is a definitive statement about a future corporate decision without any cited source or official communication from Anthropic. It's a guess dressed up as insider info.
Original quote: βanthropic is drawing a line in sand saying we will no longer push these models to be better than you know opus 5.5 you know that level of performance on these relevant areas so it'll be interesting to see if others do the sameβ
Will other labs follow suit or 'release all it out there?' β a false dilemma, as if those are the only two options. π β False Dilemma (20/100)
At 10:47
He's setting up a 'this or that' scenario for other AI labs. The world's a bit more complex than two choices, isn't it? π
Why this score: The speaker presents a false dilemma by suggesting only two possible outcomes for other AI labs: either they 'follow suit' (implying they also limit their models) or they 'release all it out there.' This ignores a spectrum of other potential strategies, such as focusing on different applications, incremental improvements, or even entirely new approaches.
Original quote: βWill other big AI labs follow suit or will they just release all it out there?β
Sources: False Dilemma Fallacy | Examples & Definition, False dilemma - Wikipedia, False Dichotomy & False Dilemma | Logical Fallacies
Kimmy K3 is 'almost like' Opus, 'twins separated at birth' β that's pure speculation, not data analysis. π€‘ β Loaded Language (45/100)
At 11:19
He's using dramatic, speculative language like 'twins separated at birth' to imply a connection that isn't proven by the chart. π©
Why this score: The speaker observes that Kimmy K3 is 'so close to Opus' on the benchmark and immediately jumps to highly speculative, loaded language like 'almost like they're the same model' and 'twins separated at birth.' While the proximity might be interesting, these phrases inject a narrative of intentional similarity or shared origin without any actual evidence beyond visual closeness on a single chart.
Original quote: βWhy is Kimmy K3 so close to Opus? So Claude so closer to the Claude family of models. It's almost like they're the same model. Could they be like twins separated at birth?β
Compares AI models to a fictional character's intelligence β a narrative, not a fact. β False Equivalence (20/100)
At 12:58
He's comparing real AI models to a fictional character's backstory. That's not data, that's a fan theory with a thesis. π
Why this score: The speaker draws a direct parallel between the 'smartest character in the whole book' (Bean from Ender's Shadow) and current AI models. While an analogy can be illustrative, presenting it as 'literally more or less what these models are' conflates a fictional narrative with the actual capabilities and development of AI. It's an emotional appeal to a story, not a technical assessment.
Original quote: βAnd that's literally more or less what these models are. They are by far the smartest models that we have access to.β
Sources: I asked ChatGPT, Claude, Gemini and Grok which sci-fi AI they're most like β and their answers were surprisingly different | TechRadar, Science Fiction AI Has Nothing to Do with Real-World βAIβ | Aaron Hertzmannβs blog, Fact or fiction? Separating the AI fantasy from reality | The Drum
Highlights that the impressive graphics are 2D, not 3D. β No Frame (75/100)
At 13:44
He's just pointing out a technical detail about the graphics. It's a clear observation. π
Why this score: The speaker is making a direct observation about the visual presentation on screen, stating 'None of this is 3D, by the way. This is all 2D graphics.' This is a factual description of what is being shown, intended to highlight the capability of the AI to generate impressive 2D visuals. There's no hidden agenda or rhetorical trick here.
Original quote: βNone of this is 3D, by the way. This is all 2D graphics and uh
that seems pretty impressive.β
Claims significant graphical elements like 3D movement and highlights were generated in an hour. β Confidence Mismatch (45/100)
At 13:56
He's claiming '3D movement' and 'highlights' were conjured in an hour, while showing 2D. Pick a lane, mortal. π
Why this score: The speaker explicitly states that 'none of this was here like an hour ago,' referring to '3D movement touchdowns highlights scores.' However, he previously stated that 'None of this is 3D, by the way. This is all 2D graphics.' This creates a contradiction or at least a significant ambiguity. If it's all 2D, then claiming '3D movement' was generated in an hour is misleading. The confidence in the rapid generation of these features, especially '3D movement,' doesn't align with the earlier clarification about the graphics being 2D.
Original quote: βthe 2D sprites are sort of like the
the character design, but the 3D
movement touchdowns highlights
scores,
none of this was here like an hour ago.β
Claiming a new camera zoom feature wasn't there before β a specific observation. β No Frame (75/100)
At 14:30
He's just pointing out a visual change he noticed. Nothing tricky here. π
Why this score: The speaker is making a direct observation about a visual element in the game, stating it wasn't present previously. This is a straightforward claim about a specific feature.
Original quote: βof this like this this whole look, the zoom in the camera. This wasn't here before.β
Comparing two Fable versions, one on a 'deeper technical' engine β setting up a comparison. β No Frame (75/100)
At 14:37
He's just explaining his setup for comparison. No spin yet, just the facts. π
Why this score: The speaker is describing his methodology for observing the game's changes, by running two versions. This is a factual setup for his subsequent analysis.
Original quote: βBasically, I opened up two versions of Fable and one of them is working on the engine behind this.β
Expressing surprise at the speed and quality of the new Fable update β a personal opinion. β Personal Story (60/100)
At 14:50
He's impressed, good for him. It's just his personal take, not a universal truth. π
Why this score: This is a subjective statement of personal opinion and experience. The speaker is sharing his reaction to the update's development, which isn't a fact-checkable claim but a personal sentiment.
Original quote: βBut I got to say, I'm kind of blown away by how quickly and how good they made it.β
Announcing Anthropic's offer of up to three resets for higher-tier plans β a specific company announcement. β No Frame (75/100)
At 15:04
A direct report of a company announcement, with conditions. This one's clean. π
Why this score: The speaker is relaying a specific announcement from Anthropic regarding resets for certain subscription tiers. This is a verifiable piece of information about a company's offering.
Original quote: βWhat's interesting is today with this release, Enthropic announced that they're going to give us, I believe it's up to three resets. It depends on how long you've been on their plan, and it's only for the higher tier plans.β
Asking for audience feedback on his 'wild theories' about Anthropic's future β inviting speculation. β Just Vibes (50/100)
At 15:18
He's just fishing for engagement on his 'wild theories.' It's a call to action, not a claim. π
Why this score: The speaker is explicitly inviting audience participation and feedback on his speculative theories. This is a conversational prompt, not a factual claim or a manipulative framing technique.
Original quote: βAnd let me know what you thought of my wild theories about what's happening with Enthropic and the future release of these models. Do you think I'm right or completely wrong?β
Hypothesizing about Anthropic's aggressive testing and potential implications for model efficiency β pure speculation. β Just Vibes (50/100)
At 15:29
He's just playing 'what if' with the audience. It's a thought experiment, not a statement of fact. π
Why this score: This entire segment is a hypothetical discussion, exploring potential scenarios and implications based on his 'theories.' It's framed as a series of questions and imaginative possibilities, not as factual assertions.
Original quote: βAnd if you think I'm right and anthropic is aggressively testing the different model architectures and doing so in this kind of autonomous way, could there be big implications for how efficient these models are at, you know, inference and training, etc. I mean, can you imagine if we find out thatβ¦β
See the full analysis with sources and timestamps β