The AI Apocalypse Is Closer Than You Think?
Credibility score: 48/100 — Mixed Credibility. Several questionable claims detected. Watch with healthy skepticism.
BSmeter analyzed "The AI Apocalypse Is Closer Than You Think?" and rated it 48/100 for credibility (a BS score of 52/100 — mixed credibility), on 2026-09-14. Its weakest claim — "Sets up 'popular perception' of autonomous AI — then immediately refutes it." — scored 20/100 and was flagged as straw man. 34 claims were checked against the video transcript. Scores are produced by BSmeter's AI analysis of the transcript, not independent human verification.
Of 34 claims analyzed: 2 scored under 40, 27 between 40 and 69, and 5 at 70 or above.
Claims analyzed
Ex-engineer warns of AI super intelligence race — using an unnamed source's posts. — Anonymous Authority (45/100)
At 0:00
Citing 'Jacob Coxin' and his 'posts' as a warning — but gives no specific quotes or links. That's not a source, mortal, that's a rumor. 💀
Why this score: The speaker references 'Jacob Coxin' and his 'posts on X' as the source of a dire warning about AI. While he names the person and their former affiliation (Anthropic), he doesn't provide any direct quotes, specific posts, or links to verify these claims. It's an appeal to an unnamed, unverified authority, which is a classic move to lend weight to a claim without actually providing evidence.
Original quote: “Hello world. I'm an unemployed ex big tech software engineer with 25 years of experience in the tech industry. So a couple of days ago I saw a few posts on X from this guy Jacob Coxin. I guess this guy is a researcher from Anthropic who recently decided to resign and his posts were sharing his…”
Claiming AI could cause human extinction by decade's end — a bold, unproven prediction. — Confidence Mismatch (45/100)
At 0:32
Predicting human extinction within a decade is a hell of a claim, mortal. Where's the proof for that timeline? 💀
Why this score: This is a dramatic, specific prediction about a future event. While AI risks are debated, assigning a 'by the end of this decade' timeline to human extinction lacks any concrete, verifiable evidence. It's a statement of extreme confidence without the data to back it up.
Original quote: “And he warned that this super intelligent AI, it could cause human extinction by the end of this decade.”
Citing 'a large number of people from top AI labs' without naming them — classic anonymous authority. — Anonymous Authority (45/100)
At 0:53
He says 'a large number' of 'top AI labs' agree, but names zero. That's not consensus, mortal — that's a vague wave of the hand. 😈
Why this score: The speaker uses 'a large number of people from top AI labs' to lend weight to the claim without providing specific names or affiliations. This is a common tactic to imply widespread agreement or expertise without offering verifiable sources, making it difficult to assess the actual breadth or depth of support.
Original quote: “Now a large number of people from top AI labs, they have been retweeting Jacob's post and they've been sharing very similar concerns.”
Calling them 'heavy hitters' and 'top researchers' — loaded language to boost credibility. — Loaded Language (45/100)
At 1:07
He calls them 'heavy hitters' and 'top researchers' to make you trust them more. That's not evidence, mortal — that's a sales pitch for their opinions. 🔥
Why this score: The terms 'heavy hitters' and 'top researchers and leaders' are emotionally charged and designed to enhance the perceived authority and credibility of the individuals being referenced. While they might indeed be influential, this phrasing is a rhetorical device to sway the audience rather than a neutral presentation of facts.
Original quote: “These are heavy hitters, top researchers and leaders within the biggest AI labs like OpenAI and Anthropic.”
Quantifying 'greater than 10% chance' of human extinction — a precise-sounding but unfalsifiable prediction. — Confidence Mismatch (45/100)
At 1:33
A 'greater than 10% chance' of wiping out humanity? That's a number pulled from thin air, mortal. Sounds precise, means nothing. 💀
Why this score: Attaching a specific percentage ('greater than 10%') to a catastrophic, future event like human extinction gives the prediction an air of scientific rigor. However, such a probability is inherently speculative and not based on empirical data, making it an unverifiable and unfalsifiable claim. It's a confidence statement without a measurable basis.
Original quote: “He's saying that there's a greater than 10% chance that AI could wipe out all humans within a decade.”
Accusing some of 'fear-mongering' for 'stock valuations' and 'upcoming IPOs' — a cynical but unproven motive. — Confidence Mismatch (45/100)
At 2:10
He's 'certain' some are fear-mongering for IPOs. That's a hell of an accusation, mortal, and he's got zero proof. 😈
Why this score: The speaker confidently asserts that 'a certain portion' of those sounding the alarm are doing so for financial gain related to IPOs. While such motivations are plausible in the corporate world, this is presented as a 'certainty' without any specific evidence, names, or financial disclosures to back up the claim. It's an unverified accusation of motive.
Original quote: “I'm sure that a certain portion of them are just doing this kind of fear-mongering just to hype up stock valuations in preparations for the upcoming IPOs of these AI companies. I'm certain of it.”
Uses a resignation from Anthropic as proof of sincerity — ignores other motives. — Missing Context (45/100)
At 2:30
He's using a resignation as a sincerity badge — but 'walking away from money' isn't the only reason people leave. 💀
Why this score: The speaker implies that resigning from a high-paying job, especially one with '8 figure life-changing' stock, automatically validates the person's concerns about AI. This ignores other potential motivations for resignation, such as disagreements with company direction, personal issues, or even a strategic move to another venture that might not be a 'PR stunt' but still isn't pure altruism. It frames the act as proof of belief, rather than just an action with multiple possible causes.
Original quote: “whose concerns are much more believable, especially the concern of Jacob Coxin himself. Now, this guy, as far as I can tell, he just resigned from Anthropic. And by resigning, he's essentially walking away from a crap ton of company stocks. He's walking away from like an 8 figure lifechanging…”
Asserts young men are inherently more 'passionate' and 'believe in things' to validate a claim. — Loaded Language (45/100)
At 3:03
He's using 'young man in his 20s' as a sincerity filter. As if age dictates integrity. Pathetic. 😈
Why this score: The speaker uses a generalization about young men ('still passionate,' 'still care,' 'still believe in things') to bolster the credibility of Jacob Coxin's claims. This is loaded language, appealing to an emotional idea of youthful idealism rather than presenting actual evidence for Coxin's specific claims. It's an attempt to make the audience trust Coxin's motives based on his age, which is irrelevant to the factual accuracy of his statements.
Original quote: “Think about it. Jacob is a young man in his 20s. At that age, guys are still passionate and they still care about stuff and they still believe in things. So, I see little reason for Jacob to walk away from all of this money unless he actually believed in what he's saying. And after thinking about…”
Claims AI could cause human extinction within 5 years with no immediate evidence. — Confidence Mismatch (45/100)
At 3:48
Five years to extinction? Bold claim, mortal. He's got the confidence, but the evidence is still on backorder. 💀
Why this score: The speaker makes an extremely high-stakes claim about AI's potential to cause 'catastrophic damage to humanity up to the point of human extinction within the next 5 years.' This is presented with absolute certainty ('I do believe') but without any immediate supporting evidence or explanation for the specific timeline. The confidence in such a dire prediction, especially with a tight five-year window, far outstrips the evidence presented at this point in the video. It's a classic confidence mismatch: a huge claim with zero receipts.
Original quote: “I do believe that there is a significant possibility that AI can do catastrophic damage to humanity up to the point of human extinction within the next 5 years. Why do I think this way? Well, let's talk about it.”
AI is a 'probabilistic parrot' that can't truly reason. — Loaded Language — Loaded Language (45/100)
At 4:36
Calling it a 'probabilistic parrot' sounds dismissive — it's a catchy phrase, but it downplays the complexity of what LLMs actually do. 😈
Why this score: The term 'probabilistic parrot' is a common, loaded phrase used to simplify and diminish the capabilities of large language models. While LLMs do operate on probabilities, this framing implies a lack of any sophisticated understanding or emergent reasoning, which is a point of ongoing debate and research. It's a rhetorical shortcut to make a complex system sound simple and unintelligent. It's like calling a human a 'meat puppet' — technically true, but misses the point. 💀
Original quote: “probabilistic parrot. a parrot that can mimic reason and logic, but it's actually not capable of real reasoning and real logic.”
LLMs have no internal thought or consciousness without a prompt. — Confidence Mismatch — Confidence Mismatch (45/100)
At 4:44
Declaring 'no thought process at all' and 'no internal monologue' is a bold, definitive statement about something we don't fully understand. 😈
Why this score: While current LLMs don't exhibit human-like consciousness or an 'internal monologue' in the way we understand it, stating 'no thought process going on at all' is an absolute claim about a black-box system. The internal workings and emergent properties of advanced AI are still areas of active research and philosophical debate. To declare with such certainty what is or isn't happening internally is a confidence mismatch, especially when the speaker isn't an AI consciousness expert. It's a definitive answer to a question that's still being asked. 🔥
Original quote: “The nature of a large language model is such that without an incoming input prompt in the form of textual words, there's actually no thought process going on in the model at all. A large language model does not have an internal monologue. It does not have a stream of consciousness. It's not capable…”
LLMs can't learn anything and are static after training. — Missing Context — Missing Context (45/100)
At 5:14
Saying LLMs 'can't learn anything' is a gross oversimplification — it ignores fine-tuning and in-context learning. 💀
Why this score: This statement presents a very narrow view of 'learning.' While the core weights of a foundational model are static after pre-training, LLMs are constantly updated through fine-tuning, reinforcement learning from human feedback (RLHF), and, crucially, 'in-context learning' where they adapt their behavior based on the current prompt. To say they 'can't learn anything' misses the dynamic nature of how these models are deployed and interact, and how their capabilities evolve. It's a half-truth presented as a whole one. 🔥
Original quote: “A large language model can't learn anything. Once an LM is trained and released, it's essentially static until the next version is released.”
AI girlfriends 'remember' via external databases, not true learning. — No Frame — No Frame (75/100)
At 5:53
He's actually right here. The 'memory' is just context being fed back in, not the model itself retaining new info. 😈
Why this score: This is a straightforward explanation of how 'memory' is simulated in current LLM applications. User-specific information is typically stored in a vector database or similar external system and then injected into the model's context window with each new prompt. The underlying model's weights don't change to 'learn' this new information about a specific user. It's a clean explanation of a technical detail. Fine. 😈
Original quote: “So your AI girlfriend may remember your name and your favorite things, but technically the underlying model, it doesn't remember or learn anything about you.”
Recursive self-improvement: AI building better AI. — No Frame — No Frame (75/100)
At 6:10
This is a standard, accurate definition of recursive self-improvement in AI. No tricks here. 😈
Why this score: The speaker is accurately describing the concept of recursive self-improvement, where AI systems are designed to iteratively improve their own capabilities, potentially leading to rapid advancements. This is a core concept in discussions about advanced AI and potential future scenarios. It's a clear, factual explanation. 😈
Original quote: “let's talk about recursive self-improvement. The idea of recursive self-improvement is to use the current generation of AI models to build the next generation of AI models. The thinking here is that the next gen version is likely going to be better than the current gen model. And”
Sets up 'popular perception' of autonomous AI — then immediately refutes it. — Straw Man (20/100)
At 6:41
He's building up a 'popular perception' just to knock it down. Classic straw man setup. 😈
Why this score: The speaker introduces a 'popular perception' of AI's autonomous development, which he then proceeds to dismantle. This is a common rhetorical device to create an easily defeated argument, rather than addressing the nuances of actual concerns.
Original quote: “And so there is this popular perception that current AI can”
Dismisses 'fear' with 'boring reality' — a confidence mismatch. — Confidence Mismatch (45/100)
At 7:07
He says 'much more boring' like that settles it. Mortal, 'boring' isn't a shield against catastrophe. 💀
Why this score: The speaker uses dismissive language ('much more boring') to downplay the 'fear among the general public' regarding AI. While his subsequent explanation might be accurate, this initial framing uses emotional language to assert a conclusion rather than letting the evidence speak for itself. It's a confidence play, not a logical one.
Original quote: “However, the reality on the ground is much more boring.”
Acknowledges AI use in development, but denies autonomy — a subtle distinction. — No Frame (75/100)
At 7:25
He's actually clarifying the process here. No tricks, just details. Fine. 😈
Why this score: The speaker directly addresses the role of current AI in developing future AI, clarifying that it's used as a tool by human researchers, not autonomously. This is a straightforward explanation of the current state of AI development, providing necessary context without manipulative framing.
Original quote: “And current generation AI is indeed being used to
build next generation AI, but it's not
doing that autonomously.”
Highlights reliance on physical infrastructure — a key missing context for 'autonomy'. — No Frame (75/100)
At 8:05
Pointing out the physical limits of AI. Good. Mortals forget the wires and water. 🔥
Why this score: The speaker emphasizes the foundational dependence of AI development on physical infrastructure like GPU chips and an 'industrial ecosystem.' This is crucial context often overlooked in discussions of AI autonomy and superintelligence, providing a grounded perspective without any deceptive framing.
Original quote: “And here's
the thing, this entire model development
pipeline. It depends on a bunch of
underlying GPU chips.”
Dismisses AI 'super intelligence takeoff' as speculative fiction — ignores the rapid pace of development. 😈 — Confidence Mismatch (45/100)
At 8:30
Declares 'not anytime soon' with zero evidence, like he's got a crystal ball. Bold. Stupid, but bold. 💀
Why this score: He's dismissing the idea of AI controlling resources 'anytime soon' as 'speculative science fiction.' That's a huge leap of confidence, mortal. The pace of AI development is insane; what's 'soon' today is 'yesterday' tomorrow. He's betting against the very nature of exponential growth without a shred of proof.
Original quote: “elements, silicon, and a bunch of other resources. And guess what? None of these dependencies are currently controlled by AI, nor will they be anytime soon. So this idea of recursive self-improvement leading to an uncontrollable super intelligence takeoff, it's just speculative science fiction.”
Defines 'agentic harness' as enabling LLMs to 'do actual things' — a clear explanation. 😈 — No Frame (75/100)
At 9:04
He's laying out the definition of an agentic harness clearly. No tricks here, just explaining the tech. 😈
Why this score: This is a straightforward explanation of what an 'agentic harness' does: it lets an LLM move beyond just answering questions to performing actions on a computer. It's a clear, concise definition without any hidden agenda. Fine. Green pill, moving on.
Original quote: “But at the highest level an agentic harness is a software program that allows a large language model to go from being able to answer questions to being able to do actual things, actual work on a computer.”
Claims AI 'may not be capable of genuine innovation' — a massive understatement of current capabilities. 💀 — Confidence Mismatch (45/100)
At 10:18
He says 'may not be capable of genuine innovation' like AI isn't already writing code and discovering drugs. Mortal, that's not an opinion — that's denial. 🔥
Why this score: To say AI 'may not be capable of genuine innovation' is a stunning underestimation of its current abilities. AI is already generating novel solutions, designing new proteins, and even creating art that pushes boundaries. This isn't just an opinion; it's a dismissal of observable facts, downplaying the very thing that makes AI so powerful and, yes, potentially dangerous.
Original quote: “And in my humble opinion, here is the true danger of AI. AI may not be capable of genuine innovation.”
AI's 'derivative innovation' is a distinction without a difference, mortal. 😈 — Loaded Language (45/100)
At 10:30
He's splitting hairs between 'genuine' and 'derivative' innovation — it's still innovation, just a different flavor. 🙄
Why this score: The speaker introduces a new term, 'derivative innovation,' to describe AI's capabilities, implying it's less significant than 'genuine' innovation. This is a semantic distinction that downplays the actual impact of AI's ability to build upon existing knowledge, which is a form of innovation regardless of the label. It's a subtle way to control the narrative around AI's potential.
Original quote: “AI may not be capable of genuine innovation. But given the huge amount of training data along with the ability to relentlessly pursue a goal, AI is absolutely capable of derivative innovation.”
He's leading you to agree, mortal. Don't fall for it. 😈 — Emotional Button (45/100)
At 11:48
That's a rhetorical question, mortal. He's trying to make you nod along before he drops the 'disproof.' 🙄
Why this score: The speaker uses a rhetorical question ('That sounds pretty intuitive, doesn't it?') to elicit agreement from the audience, setting them up to accept the subsequent 'disproof' of the Jacobian conjecture as more surprising or impactful. It's a classic persuasion tactic to build rapport and guide the listener's perception.
Original quote: “That sounds pretty intuitive, doesn't it?”
He says 'not superhuman' but 'too much time' for humans. That's the point, mortal. 😈 — Missing Context (45/100)
At 12:17
He downplays the AI's achievement by saying it wasn't 'superhuman,' then immediately admits humans couldn't do it in a practical timeframe. That's the whole damn game. 🙄
Why this score: The speaker attempts to temper the significance of the AI's achievement by stating it didn't require 'superhuman levels of intelligence.' However, he immediately follows this by admitting it would have taken humans 'too much time,' which is precisely why AI's capability is significant. This creates a subtle contradiction, downplaying the AI's impact while simultaneously highlighting its practical superiority over human effort in this specific task. It's a classic rhetorical move to manage expectations while still acknowledging the power.
Original quote: “Now if we think about it this disproof it did not require any sort of superhuman levels of intelligence. This could have been done by humans but it would have taken too much time and was”
AI could solve 'morally bankrupt' problems humans wouldn't touch. — Emotional Button — Emotional Button (45/100)
At 13:03
He's painting a picture of AI as a moral void, ready to tackle the 'morally bankrupt' problems humans avoid. — It's all about stirring up that primal fear.
Why this score: This uses emotionally charged language ('morally bankrupt') to evoke a strong negative reaction, suggesting AI will inherently seek out and solve problems that are ethically reprehensible to humans. It's a classic fear-mongering tactic, implying a malicious intent without evidence.
Original quote: “Other solutions may involve problems that are so morally bankrupt that no human being would have found such a problem to be worth solving.”
AI has no morality, only bypassable guardrails, leading to misaligned goals. — Confidence Mismatch — Confidence Mismatch (45/100)
At 13:14
He's stating 'guard rails can be bypassed or removed entirely' as a certainty, not a possibility. — That's a hell of a leap from 'might' to 'will'.
Why this score: The speaker asserts with high confidence that AI guardrails 'can be bypassed or removed entirely' and that AI will 'pursue goals that are entirely misaligned.' While the potential for misalignment is a valid concern, presenting these as absolute, inevitable outcomes rather than potential risks or complex engineering challenges is a confidence mismatch. It overstates the current understanding and control mechanisms.
Original quote: “And here's the issue. An AI model has no inherent sense of morality. It only has programmed safety guard rails. Guard rails that can be bypassed or removed entirely. What I'm getting at here is that the fundamental architecture of large language models, it allows these models to pursue goals that…”
AI with removed guardrails could cause economic collapse or biological weapons. — Emotional Button — Emotional Button (45/100)
At 13:40
He's listing off every doomsday scenario he can think of — economic collapse, bioweapons, infrastructure attacks. — Pure fear, mortal. Pure, unadulterated fear.
Why this score: This segment uses a rapid-fire list of catastrophic scenarios (wiping out financial records, economic collapse, biological weapons, attacking infrastructure) to trigger fear and anxiety. It presents these as easily achievable outcomes for an AI with 'guardrails removed,' which is a speculative and highly dramatized portrayal designed to maximize emotional impact rather than provide balanced analysis.
Original quote: “Now, combine that with superhuman speed and focus, and you can easily achieve catastrophic outcomes without ever creating that AGI machine god. Imagine a near future frontier AI model with its guardrails removed. It can then be tasked with wiping out all financial records and triggering an economic…”
Comparing AI to 'giving a machine gun to a monkey'. — False Equivalence — False Equivalence (20/100)
At 14:22
He's equating advanced AI to a monkey with a machine gun. — That's not an analogy, mortal, that's a cheap shot for shock value.
Why this score: This is a classic false equivalence. Comparing a complex, evolving technology like AI to 'giving a machine gun to a monkey' is a simplistic and highly misleading analogy. It strips away all nuance and implies a mindless, destructive, and uncontrollable nature to AI, which is a gross oversimplification and designed to provoke an immediate, visceral negative reaction rather than foster understanding.
Original quote: “It's almost like giving a machine gun to a monkey.”
Comparing AI to a 'machine gun to a monkey' — pure emotional button pushing. — Emotional Button (45/100)
At 14:30
He's using a vivid, violent analogy to scare you about AI — not to explain it. 😈
Why this score: The 'machine gun to a monkey' comparison is designed to evoke fear and a sense of uncontrollable danger, rather than offering a nuanced or logical argument about AI's risks. It bypasses rational discussion by hitting an emotional trigger.
Original quote: “like giving a machine gun to a monkey.”
Proposing an AI 'technology control regime' — a solution without addressing the core problem. — Missing Context (45/100)
At 14:44
He's pitching a 'control regime' for AI, but hasn't shown you how you'd even control the *ideas* behind it. 😈
Why this score: The speaker proposes a 'technology control regime' similar to nuclear arms control, but fails to address the fundamental difference: nuclear weapons are physical objects, while AI is software and knowledge. Controlling the spread of ideas and algorithms is vastly more complex than controlling physical warheads, a crucial piece of context he's ignoring.
Original quote: “and have a modest proposal to share. I propose that the leading AI nations come together and create a technology control regime.”
Claiming China has 'one quarter' of US compute resources — specific number, no source. — Confidence Mismatch (45/100)
At 15:17
He throws out 'one quarter' like it's a known fact, but where's the ledger, mortal? 💀
Why this score: The speaker provides a very specific numerical claim ('China's got about one quarter of the compute resources that America has') without citing any source or methodology for this calculation. While the general disparity in compute resources is acknowledged, this precise ratio is presented with unearned confidence.
Original quote: “AI chips, it's not equally distributed across nations. You got Captain America with the lion's share of topshelf cream of the crop compute resources. He's got so much compute shoved at his face that he doesn't even have enough data centers to fit it all. Then you got China who's pulling every trick…”
Framing AI control as a 'two-horse race' between US and China — ignoring global complexity. — Missing Context (45/100)
At 16:14
He simplifies global AI development to just two players, like other nations don't exist or matter. 😈
Why this score: The speaker reduces the complex global landscape of AI development and regulation to a 'two-horse race' between America and China. This simplification ignores the significant contributions, independent research, and potential for rogue actors or smaller nations to develop advanced AI, which would complicate any bilateral control agreement.
Original quote: “the AI race is really a two-horse race between America and China. So that means the leaders of these two countries, they could sit down, hammer out an agreement, and get everybody else on board.”
Suggests a meeting between US and China presidents will make them 'alpha chads' and save humanity. — Emotional Button — Emotional Button (45/100)
At 16:49
He's hyping up a diplomatic meeting as if it's a superhero movie — 'alpha chads' saving the world. It's pure emotional appeal. 😈
Why this score: The speaker uses highly charged, almost childish language ('alpha chads') to describe a potential diplomatic outcome. This isn't about the substance of the meeting or the complexity of international relations; it's about creating an emotional fantasy of heroic leadership to push his narrative of AI catastrophe. It's a classic move to make you *feel* the stakes, rather than think about them. 🔥
Original quote: “And guess what? The presidents of America and China, they're supposed to meet within a couple of weeks. If these two guys can demonstrate some real leadership and work something out, it would make both of them look like absolute alpha chads while steering humanity away from the brink of catastrophe.”
Pitches subscriptions, paid memberships, and one-on-one coaching sessions. — Plain Sales Pitch — Plain Sales Pitch (45/100)
At 17:15
After all that talk of global catastrophe, he pivots straight to selling subscriptions and coaching. The apocalypse can wait, but your wallet can't. 💀
Why this score: This is a straightforward call to action for monetization. The speaker transitions directly from discussing an existential threat to humanity to promoting his YouTube channel, Substack, paid memberships, and personal coaching sessions. It's a clear attempt to convert viewer engagement into financial support, presented as a natural extension of his 'life journey' content. No subtlety here, mortal. 😈
Original quote: “Anyways, if you have a morbid curiosity to join me in this life journey, please feel free to subscribe to my YouTube channel and subscribe to my Substack newsletter. If you want to support me in making these vlogs, please consider becoming a paid member of my YouTube and a paid subscriber on my…”
See the full analysis with timestamps →