Unsupervised Learning · 2025-11-22 · 7 min
Key moments - from our scoring
Substance score
36 / 100
Five dimensions, 20 points each
The episode challenges the philosophical debate over whether AI truly understands or possesses intelligence by proposing a pragmatic alternative: evaluate AI capabilities based on ground-truth outputs rather than mechanism introspection. The host uses a compelling case study - an AI-generated blues rendition of Without Me that produces emotionally moving, novel music - to illustrate that if a human created output of equivalent quality, we would unhesitatingly attribute it to intelligence and understanding. This reframes the entire conversation away from black-box opacity toward observable results. The argument applies across domains: if AI models can replace human workers, accomplish complex goals, or produce work indistinguishable from human-created content in quality, then by definition they demonstrate understanding. The host acknowledges that neither human brains nor neural networks yield obvious "loci" of understanding when examined mechanistically, yet we don't deny human cognition on those grounds. The episode draws a sharp distinction between the universal opacity of emergence as a phenomenon and the specific implementation details of any intelligence stack, arguing that conflating these two issues has led to circular, unproductive debate. This framing benefits founders, technologists, and operators making investment or deployment decisions about AI, as it shifts focus from unanswerable mechanistic questions to measurable capability assessment.
Judge AI by its output and whether producing that output would require intelligence if a human did it. If AI produces that same output, then intelligence was used to produce it, regardless of whether we can explain the mechanism.
We cannot locate understanding or intelligence inside the human brain either - we have no idea how memories are stored or thoughts generated - yet we don't deny human intelligence because we observe intelligent outputs. The same logic should apply to AI.
An AI-generated blues cover of Eminem's 'Without Me' that creates a novel, emotionally moving piece of music that would be considered intelligent if produced by a human musician.
No; the opacity of emergence is universal across human brains, animal cognition, and AI systems. It's a general limitation of current science, not a specific flaw of AI that disqualifies it from possessing intelligence.
Our reviewer’s read on each dimension, with quotes from the episode.
The episode offers a coherent reframing of AI evaluation - judging by output rather than mechanism - that is moderately useful but largely relies on a single core argument repeated several times. The core thesis is relatively straightforward and doesn't accumulate dense clusters of novel operational insights; instead, it circles around the same philosophical point with musical and medical examples.
Judge capabilities by their ground truth outputs. In other words, in your lexicon, did the creation of that output require understanding and or intelligence if it were a human doing it?
we can start from ground truth, which is what we already know and accept as being the product of intelligence
The argument itself - that we should evaluate AI by outputs rather than internal mechanism - is a reasonable reframing, but it's not particularly contrarian or first-principles; it echoes existing pragmatist philosophy in AI (Searle's work, etc.). The analogy to human brain opacity is sound but well-trodden. The episode doesn't introduce novel data, frameworks, or counterintuitive claims beyond this single axis.
Judge capabilities by their ground truth outputs.
we still lack transparency into emergence itself, not just for tech, not just for llms, not just for AI, but for humans and other animals as well
This is a solo monologue with no guest. The host appears to be the only speaker, eliminating any opportunity to evaluate guest credibility or operational track record.
I mean, have you listened to something that I think captures extraordinarily well?
The episode uses two concrete examples - a blues version of Eminem's 'Without Me' and a vague reference to 'AI digital workers' - but offers minimal concrete data, metrics, or named cases. The musical example is illustrative but not quantified; no companies, models, benchmarks, or numbers are provided to ground the claims.
This is a blues version of Without Me by Eminem. It's from the 1950s, which means it's not real.
If AI models and scaffolding can be assembled into a product that can replace human workers, it's intelligent
This is an uninterrupted monologue with no guest, no adversarial push-back, and no follow-up questions. There is no conversational craft to evaluate; the host simply articulates a position without challenge or dialogue.
I mean, have you listened to something that I think captures extraordinarily well?
So let's listen to it.
Computed from the transcript - who did the talking, and the words that came up most.
How we can use an output-based system to judge whether or not different kinds of technology achieve understanding or intelligence. Become a Member: See omnystudio.com/listener for privacy information.
Transcribed and scored by The B2B Podcast Index.
WEBVTT - Judge AI based on Output, Not Mechanism I mean, have you listened to something that I think captures extraordinarily well? Why? Arguments that AI don't understand anything and can't possibly understand anything are completely misguided and empty. This is a blues version of Without Me by Eminem.
It's from the 1950s, which means it's not real. And he's also never done a blues version of Without Me, to my knowledge. And so it's AI generated and it's objectively a stunning piece of music, and it's quite different from the original. So let's listen to it.
Guess who's been back again? Shady's back. Tell a friend. Guess who's back?
Guess who's back? Guess who's back. Guess who's back. Guess who's back.
Guess who's back. Guess who's back. Guess who's back. Guess who's back.
Guess who's back. No no no no no no no no. I've created a monster. Cause nobody wants to see my shoes no more.
They won't shake I'm chopped liver. Well, if you want shady. This is what I give you. A little bit of me mixed with some hard liquor.
Some vodka that'll jumpstart my heart quicker than a shark. When I get shot at the hospital by the doctor. When I'm not cooperating, when I'm rocking the table while it's operating. Hey, you waited this long to stop debating cause I'm back.
I'm on the. I know that you got a job, Miss Cheney, but your husband's heart problems complicating. So the FCC won't let me be. Or let me be me.
So let me see. They try to shove me down on MTV, but it feels so empty without me. So every time I listen to that, I feel compelled to move. I think if music makes you dance and feel things, it is real.
If AI models and scaffolding can be assembled into a product that can replace human workers, it's intelligent, i.e. it has the ability to understand, pursue, and accomplish goals. If a technology can perform a task and produce an output that requires understanding, it understands.
So in this frame, understanding is the ability of an actor to interpret a given task and desired outcome well enough to create an acceptable result. AI can clearly do that now across so many domains. It's true that if you break open a neural net or a human brain and start poking at it with a stick or a scalpel or an electron microscope. There is no place to point to and say this is understanding, or here is the intelligence, but it is there in both human brains and in neural nets, because we see the outputs that prove that it's there.
We should stop wasting cycles on does it understand or is it intelligent or it can't be intelligent, because all these behaviors in both animals and technology are the result of emergent functionality. And the core issue here is that we still lack transparency into emergence itself, not just for tech, not just for llms, not just for AI, but for humans and other animals as well. So let's not confuse that opacity of emergence itself, which is a universal human problem in curiosity, with a specific implementation of that emergence, opacity and a new intelligence stack judge capabilities by their ground truth outputs.
In other words, in your lexicon, did the creation of that output require understanding and or intelligence if it were a human doing it? And if so, then did a non-human actually produce that? Did a non-human technology produce that same thing that if you saw it from someone else, it would have required intelligence? Then guess what that is?
Intelligence. Intelligence was used to produce the output. We can use the output itself, and the fact that we have defined it as requiring intelligence to say that anything that could have produced it had intelligence itself. I think this framing helps clarify the whole situation a little bit because we can start from ground truth, which is what we already know and accept as being the product of intelligence, right?
If you hear a song like this, if you see a work output from an AI digital worker or something, and you say, well, if a human would have made that, I would have thought it was a good product. I would have thought this definitely required intelligence. That statement there we can use as ground truth. And then from there, it's a quick step to say anything that can produce that then also has that intelligence.
And notice that this is completely separate from being able to explain how it got it. We just have to remind ourselves we don't know how we got ours either. We have no idea how. When you look at a spongy pink brain, how you can store memories in there, how you can have ideas, how you can have thoughts.
We have no idea where inside of that brain any of this stuff is actually performed or stored. Now, in humans, we are not tempted to say, well, since I can't find it, we are clearly not doing understanding. We are not doing intelligence. Those things are not there because I cannot find them by looking at the substrate.
We're not tempted to say that with humans, and we're not tempted to say it, because we can actually look at the outputs of ourselves doing those exact things. So why are we making this mistake with a different type of intelligence? Why are we looking at outputs that we would judge as being intelligent or requiring intelligence to make and saying, well, because I can't find where it was made or how it was made, it must not be intelligence. It just doesn't make sense.
And hopefully this frame will help you have the conversation with yourself or with others. We'll see you in the next one.
Other episodes covering the same guests and topics, from across The B2B Podcast Index.
Unlocking Venture Growth Equity in AI: Al Tarar and Rizwan Muhammad of Quartus Capital PartnersATLalts · on Neural networks85 / 100
Jack Hidary, CEO of Sandbox AQ | The Third Quantum RevolutionThe BreakLine Arena · on Neural networks83 / 100
DOP 356: Warehouse Robots Are a Distributed SystemDevOps Paradox · on Neural networks83 / 100
On Device AI vs Cloud AI, How Sensory Built 30+ Years of Voice Innovation with Todd MozerBuilt to Scale: B2B Growth with Rym Benchaar · on Neural networks83 / 100
A Conversation about Designing Human-AI Collaboration PlaybooksArticle Audio · on Neural networks80 / 100
The Evolution and Impact of AI and Machine Learning Across IndustriesBeyond the Screen · on Neural networks79 / 100AI Predicts the Text of Answers52 / 100Most Companies Aren't Anywhere Near Ready for AI61 / 100We're All Building a Single Digital Assistant67 / 100Why AI Will Replace Knowledge Workers66 / 100Why I Believe in SOTA Models Over Custom Ones49 / 100