748K views · Mar 16, 2016 · 2:12 · 1 HN recommender · 4 mentions · discussed 2021–2025
A two-minute BBC Horizon clip demonstrating the McGurk effect, where what you see of a speaker's mouth changes the sound you hear.
Why Hacker News points people here: Recommended as a quick demo of vision overriding hearing, and cited as an example of perception producing confident wrong answers.
“Very worth your two minutes if you're not yet familiar with the effect”HN comment, 2021
“Very worth your two minutes if you're not yet familiar with the effect”
Two-minute demonstration of the McGurk effect.
Are you aware that the same exact uncompressed recording sounds different depending on context? This is known as the McGurk effect. Very worth your two minutes if you're not yet familiar with the effect: this video
“The visual processing part of the brain can completely overwhelm auditory processing”
Demonstrates vision overwhelming auditory processing.
The visual processing part of the brain can completely overwhelm auditory processing: see the mcgurk effect. this video I am not at all surprised that there are also olifactory illusions which are impossible to “unsee”.
“The McGurk effect”
Example of unconscious lip reading changing what we hear.
The McGurk effect this video is another example, which might be described as people learning to lip read without them being aware that they have done so.
“see the McGurk Effect”
Example of sensory systems producing consistent wrong interpretations.
This is, I think, a better way to think about LLM mistakes compared to the usual "hallucinations". I think of them as similar to human optical illusions. There are things about the human visual cortex (and also other sensory systems, see the McGurk Effect [0]), that, when presented with certain kinds of inputs, will consistently produce wrong interpretations/outputs. Even when we are 100% ware of the issue, we can't prevent our brains from generating the incorrect interpretation. LLMs seem to have similar issues along dramatically different axes, axes that humans are not used to seeing these kinds of mistakes; where nearly no human would make this kind of mistake and so we interpret it (in my opinion incorrectly) as lack of ability or intelligence. Because these are engineered systems, we may figure out ways to solve these problems (although I personally think the best we will ever do is decrease their prevalence), but more important is probably learning to recognize the places that LLMs are likely to make these errors, and, as your comment suggests, design work flows and systems that can deal with them. [0] this video
