Chat GPT Makes up stuff
- Thread starter Rick
- Start date
Hodgo
Experienced Member
- Messages
- 1,298
He’s not showing ChatGPTs gaps, he’s showing the gap between what he thinks it is and what it actually is. Using the voice function and flooding the context window with unnecessary chatter already put him a leg behind, but the core point is he is not approaching this like a serious researcher and the result is partly due to his input, partly on the model.
Here’s two minutes of me promoting Claude’s Opus 4.7 via chat, not voice:
What are the two most influential (economically and socially) guitar pedals?
12:13 PM
Weighed competing candidates across economic and cultural impact
There's no objective answer here, but the strongest case is probably for the Maestro FZ-1 Fuzz-Tone and the Ibanez Tube Screamer.
The Fuzz-Tone, released by Gibson's Maestro division in 1962, was the first mass-market guitar effects pedal. It sat in warehouses barely selling until 1965, when Keith Richards used it on the riff for "(I Can't Get No) Satisfaction" because his fuzz-driven scratch take was meant to be replaced by horns later. The single went to #1, the warehouses emptied within months, and the entire effects pedal industry — now worth hundreds of millions annually — basically traces its commercial existence back to that moment. Socially, it redefined what an electric guitar was allowed to sound like; before fuzz, dirty tone was an accident, after fuzz it was a deliberate artistic choice.
The Tube Screamer (TS808 in 1979, TS9 in 1982) is the other pillar. Stevie Ray Vaughan made it the defining sound of '80s blues revival, but its real influence is structural: it's the most-cloned pedal in history. The entire boutique pedal economy — thousands of small builders, the whole "pedalboard culture" of the last 25 years — is built on Tube Screamer variants and the JRC4558 chip mythology around it. It also became the secret weapon for metal (used as a clean boost into high-gain amps by basically everyone from Metallica to modern djent), so its reach across genres is unusually broad.
Honorable mentions where you could reasonably swap one out: the Dallas Arbiter Fuzz Face (Hendrix), the Cry Baby wah (in continuous production since 1966), and the Boss DS-1 (best-selling distortion pedal ever, Cobain's main pedal).
12:14 PM
The origin is one of the great happy accidents in music history. In 1960, Snoddy was engineering a session for Marty Robbins's song "Don't Worry" when a faulty channel in the mixing console caused bassist Grady Martin's six-string bass to come through with a distorted, fuzzy growl. Rather than re-record the part, they kept it — and the song became a hit, with people calling the studio asking how to get that sound.
Snoddy and Hobbs then reverse-engineered the effect into a transistor-based circuit and sold the design to Gibson, who released it through their Maestro brand in 1962 as the FZ-1 Fuzz-Tone. Snoddy reportedly received a modest royalty arrangement rather than a windfall, though he lived to see his accidental invention reshape popular music. He passed away in 2018 at age 96.
12:14 PM
Unearthed source material distinguishing Hobbs's technical contributions from Snoddy's conceptual role
Unearthed source material distinguishing Hobbs's technical contributions from Snoddy's conceptual role
The answer most enthusiasts in the know would give you: Hobbs did more of the actual circuit design lift, Snoddy did the everything-else lift.
The clearest account comes from Revis Hobbs's son Robert, who tagged along to the workshop sessions as a kid. The split looked roughly like this:
Snoddy owned the origin and the hustle. He engineered the original Marty Robbins session where the broken console produced the fuzz, recognized there was money in bottling that sound, recruited Hobbs, drove the finished prototype to Gibson's Maestro headquarters in Chicago, and made the pitch. He's also the named patentee on US Patent #3,213,181.
Hobbs owned the soldering iron. He brought specialized germanium transistors Snoddy didn't have, and over roughly eight months of work split between their two workshops, he's the one who actually finished the three-transistor clipping circuit that turned a sine wave into a square wave. By the most detailed account, Hobbs is credited as the one who "finished the world's first three-transistor fuzz circuit."
The reason this gets muddled in most retellings is personality-driven: Snoddy was outgoing and gave all the interviews over the decades, often without mentioning Hobbs at all. Hobbs was reportedly a quiet, head-down engineer who actively avoided the spotlight. So the historical record skews heavily toward Snoddy's voice, but the substance suggests Hobbs was the one who solved the actual electrical engineering problem.
A fair summary: Snoddy was the producer, Hobbs was the engineer — and like a lot of producer/engineer credits, the producer got the fame.
Here’s two minutes of me promoting Claude’s Opus 4.7 via chat, not voice:
What are the two most influential (economically and socially) guitar pedals?
12:13 PM
Claude responded: There's no objective answer here, but the strongest case is probably for the Maestro FZ-1 Fuzz-Tone and the Ibanez Tube Screamer.
Weighed competing candidates across economic and cultural impactWeighed competing candidates across economic and cultural impact
There's no objective answer here, but the strongest case is probably for the Maestro FZ-1 Fuzz-Tone and the Ibanez Tube Screamer.
The Fuzz-Tone, released by Gibson's Maestro division in 1962, was the first mass-market guitar effects pedal. It sat in warehouses barely selling until 1965, when Keith Richards used it on the riff for "(I Can't Get No) Satisfaction" because his fuzz-driven scratch take was meant to be replaced by horns later. The single went to #1, the warehouses emptied within months, and the entire effects pedal industry — now worth hundreds of millions annually — basically traces its commercial existence back to that moment. Socially, it redefined what an electric guitar was allowed to sound like; before fuzz, dirty tone was an accident, after fuzz it was a deliberate artistic choice.
The Tube Screamer (TS808 in 1979, TS9 in 1982) is the other pillar. Stevie Ray Vaughan made it the defining sound of '80s blues revival, but its real influence is structural: it's the most-cloned pedal in history. The entire boutique pedal economy — thousands of small builders, the whole "pedalboard culture" of the last 25 years — is built on Tube Screamer variants and the JRC4558 chip mythology around it. It also became the secret weapon for metal (used as a clean boost into high-gain amps by basically everyone from Metallica to modern djent), so its reach across genres is unusually broad.
Honorable mentions where you could reasonably swap one out: the Dallas Arbiter Fuzz Face (Hendrix), the Cry Baby wah (in continuous production since 1966), and the Boss DS-1 (best-selling distortion pedal ever, Cobain's main pedal).
You said: Who invented the Maestro?
Who invented the Maestro?12:14 PM
Claude responded: The Maestro Fuzz-Tone (FZ-1) was invented by Glenn Snoddy, a Nashville recording engineer, along with his colleague Revis Hobbs (a WSM radio engineer).
The Maestro Fuzz-Tone (FZ-1) was invented by Glenn Snoddy, a Nashville recording engineer, along with his colleague Revis Hobbs (a WSM radio engineer).The origin is one of the great happy accidents in music history. In 1960, Snoddy was engineering a session for Marty Robbins's song "Don't Worry" when a faulty channel in the mixing console caused bassist Grady Martin's six-string bass to come through with a distorted, fuzzy growl. Rather than re-record the part, they kept it — and the song became a hit, with people calling the studio asking how to get that sound.
Snoddy and Hobbs then reverse-engineered the effect into a transistor-based circuit and sold the design to Gibson, who released it through their Maestro brand in 1962 as the FZ-1 Fuzz-Tone. Snoddy reportedly received a modest royalty arrangement rather than a windfall, though he lived to see his accidental invention reshape popular music. He passed away in 2018 at age 96.
You said: Did Snoddy or Hobbs do more of the lift?
Did Snoddy or Hobbs do more of the lift?12:14 PM
Claude responded: Honestly, the historical record is thin enough on this that I'd be making things up if I gave you a confident split.
Honestly, the historical record is thin enough on this that I'd be making things up if I gave you a confident split. Let me check what's actually documented.Unearthed source material distinguishing Hobbs's technical contributions from Snoddy's conceptual role
Unearthed source material distinguishing Hobbs's technical contributions from Snoddy's conceptual role
The answer most enthusiasts in the know would give you: Hobbs did more of the actual circuit design lift, Snoddy did the everything-else lift.
The clearest account comes from Revis Hobbs's son Robert, who tagged along to the workshop sessions as a kid. The split looked roughly like this:
Snoddy owned the origin and the hustle. He engineered the original Marty Robbins session where the broken console produced the fuzz, recognized there was money in bottling that sound, recruited Hobbs, drove the finished prototype to Gibson's Maestro headquarters in Chicago, and made the pitch. He's also the named patentee on US Patent #3,213,181.
Hobbs owned the soldering iron. He brought specialized germanium transistors Snoddy didn't have, and over roughly eight months of work split between their two workshops, he's the one who actually finished the three-transistor clipping circuit that turned a sine wave into a square wave. By the most detailed account, Hobbs is credited as the one who "finished the world's first three-transistor fuzz circuit."
The reason this gets muddled in most retellings is personality-driven: Snoddy was outgoing and gave all the interviews over the decades, often without mentioning Hobbs at all. Hobbs was reportedly a quiet, head-down engineer who actively avoided the spotlight. So the historical record skews heavily toward Snoddy's voice, but the substance suggests Hobbs was the one who solved the actual electrical engineering problem.
A fair summary: Snoddy was the producer, Hobbs was the engineer — and like a lot of producer/engineer credits, the producer got the fame.
Strat Avenger
Hero Member
- Messages
- 3,040
It does make stuff up. I have seen it regarding other topics.
Hodgo
Experienced Member
- Messages
- 1,298
Of course it does, that’s what it’s designed to do: use stochastic modeling to determine the most likely token (chunk of a word or whole word) based on the context it has. It is not thinking. It is not error checking. The quality of the input (training data + prompt + model) determines the output. If you trained an LLM only on patent data and canonical sources, it would be better. Even if you asked a generic model to only reference those sources, the quality would improve.
What’s going on in this video is that A) he’s junking up the context with chatter B) he’s using the voice mode which is further junk as the LLM attempts to keep it conversational (hence why it didn’t show him sources). As someone who uses AI daily, it’s like watching someone use a step bit chocked into an impact driver to drill tuner holes… wrong tool used for the wrong desired end state, but it might get close enough for government work.
stratamania
Mythical Status
- Messages
- 12,433
I am not even going to waste my time watching the clickbait video.
Similar threads
