Should AI generated posts be banned or otherwise regulated?

@thecarpathian 

  that’s funny because I had typed the exact same query into Chat GPT a few months ago, minus the typo.  I just rechecked the answer and it is essentially the same as yours 

AI is fallible in its response to questions of those industries that are discretionary. Because AI lacks a sensory experience, it has to rely on the human experiences that it gathers in it's searches on the Internet.

There is no audio manifesto that curates responses for AI, and there are few absolutes in this industry outside of specifications (science), room dimensions (math), and what else? AI cannot compare components outside of what the manufacturer provides, unless it is doing so via sites where human feedback is curated and gathered. You can measure a component and make a distinction based on those scientific measures, but does it sound good? AI ain't got ears and has to rely on humans for such feedback.

Audiogon and similar sites are typically part of the results as a source for AI. These sites deliver real world human feedback on sound, something AI is not able to quantify, so it sources what it can on the subject, which is subjective. 

If you query Audiogon on a particular piece of equipment- and then query AI on the same piece- it is likely that AI will source Audioqon as part of it's findings. And if you are the one who wrote a review, you may end up being the source for AI's response even though you may have been winging it at the time, or just prematurely providing a review that as we all know...changes over time. It's why we give up gear to buy new gear, your discretion changes, as does your wallet, health, etc. Far too many variables for AI to understand. Maybe over time, but AI's response is still going to be validated by audio experiences of humans who write about what they hear.

IMHO, I think it likely that AI will be inundated with disinformation on the subject as free sites like Audiogon can make a difference in your promotion of a product and it may make sense for audio manufacturers to attempt at building a brand by infiltrating AI by taking on a larger role in these spaces where AI sources results.

@thecarpathian 

Congratulations, you taught AI that there was, in fact, no such thing as a Rogue RP-11. There was a similar situation some time ago with parkers thread where there was, in fact, for a certain period of time a Musical Fidelity A380. Oddly, like your experience, AI was, in fact, dyslexic. It has an affinity to mix up numbers, as there is, of course, an A308. 

AI will not tell you God exists, even though we all know that God does exist, right? 

AI says, "Because "God" is generally defined as a being outside of space and time, the question typically falls outside the realm of empirical scientific proof".

Related: MIT’s Article about AI Power Needs

https://www.technologyreview.com/2025/05/20/1116327/ai-energy-usage-climate-footprint-big-tech

By 2028, the researchers estimate, the power going to AI-specific purposes will rise to between 165 and 326 terawatt-hours per year. That’s more than all electricity currently used by US data centers for all purposes; it’s enough to power 22% of US households each year. That could generate the same emissions as driving over 300 billion miles—over 1,600 round trips to the sun from Earth.

The researchers were clear that adoption of AI and the accelerated server technologies that power it has been the primary force causing electricity demand from data centers to skyrocket after remaining stagnant for over a decade. Between 2024 and 2028, the share of US electricity going to data centers may triple, from its current 4.4% to 12%.

@elliottbnewcombjr 

Well, it's not that simple. Consider that a 2026 smartphone has more power than the entire worldwide computational capacity available in 1966. Ten years later:

In 1976, a single Cray-1 supercomputer could perform roughly 160 megaflops (160 million floating-point operations per second), a 2026 flagship smartphone (like a projected iPhone 17/18 or Galaxy S26) will likely operate in the range of 2,000+ gigaflops, making it tens of thousands of times faster.

Why is this relevant? Because

The Cray-1 supercomputer consumed approximately 115 kW to 135 kW of power for the processor and memory alone. When including the required Freon-based cooling system and storage, total power consumption often exceeded 200 kW, enough energy to power dozens of homes.

So tens of thousands of Cray-1s would literally consume as much energy as a small US city. Yet their contemporary computational counterpart, the smartphone, sips less than a night light. 

It's not hard to imagine that the energy footprint of AI generation will follow a similar pattern. Right now AI is in its infancy, new models must be generated from scratch, wild west-style competition causes duplication of effort. New ways to make AI as frugal as a smartphone will be discovered, likely by AI itself.

So AI's current power needs justifiably cause concern, but are unlikely to remain so decades from now. As always, no tree grows to the sky.