What are your thoughts about/Interactions with AI?

[jlhoots]

So glad to be retired!!

[Stompin at the Savoy]

One area where AI could potentially be very useful: debunking misinformation and viral fallacies on the web, social media, etc.

Edited Friday at 10:24 PM by Stompin at the Savoy

[Kevin Bresnahan]

6 hours ago, jlhoots said:

So glad to be retired!!

My thoughts exactly!!

[jazzbo]

Me too, for 17 years now.

[Kevin Bresnahan]

15 hours ago, Stompin at the Savoy said:

One area where AI could potentially be very useful: debunking misinformation and viral fallacies on the web, social media, etc.

Since AI likely creates a majority of them, it would be a bit self-defeating.

[Stompin at the Savoy]

1 hour ago, Kevin Bresnahan said:

Since AI likely creates a majority of them, it would be a bit self-defeating.

You are probably right that AI is used to create a majority of fake videos but it is not behind most misinformation or fallacious reports.  Let me give an example: recently I saw a meme on Facebook which falsely quoted the Chinese philosopher Chuang Tzu.  Now in Chuang Tzu there is a story about a man dreaming he was a butterfly.  But this meme which claimed to be a quote said a caterpillar thought its life was all over just before its metamorphosis.  OK, so this is just false:  there is no Chuang Tzu quote about a caterpillar.  Dozens of people had shared this nonsense to their own FB pages.  AI, which is to a great extent a creature of the web, could quite easily be used to detect this sort of bs.  For example a web browser could have an extension which checks the extent of confirming documents.  If this meme or article or variants of it is the only source for a claim, the extension could put a warning on it.  Things that have already been extensively debunked by reliable sources could be marked, etc.

Another example.  A friend shared an article on FB which described Admiral Holsey calling out Trump at a meeting in front of various military bigwigs and Trump walking out.  I knew that was fake and notified my friend because I read national papers every day and nobody had ever mentioned this business.  When I checked, several places had debunked it too.  Detecting this sort of thing is very easy to do with AI, imo.

Even the fake videos could probably be quite easily detected and marked by AI.

Edited Saturday at 03:02 PM by Stompin at the Savoy

[Kevin Bresnahan]

1 hour ago, Stompin at the Savoy said:

You are probably right that AI is used to create a majority of fake videos but it is not behind most misinformation or fallacious reports.  Let me give an example: recently I saw a meme on Facebook which falsely quoted the Chinese philosopher Chuang Tzu.  Now in Chuang Tzu there is a story about a man dreaming he was a butterfly.  But this meme which claimed to be a quote said a caterpillar thought its life was all over just before its metamorphosis.  OK, so this is just false:  there is no Chuang Tzu quote about a caterpillar.  Dozens of people had shared this nonsense to their own FB pages.  AI, which is to a great extent a creature of the web, could quite easily be used to detect this sort of bs.  For example a web browser could have an extension which checks the extent of confirming documents.  If this meme or article or variants of it is the only source for a claim, the extension could put a warning on it.  Things that have already been extensively debunked by reliable sources could be marked, etc.

Even the fake videos could probably be quite easily detected and marked by AI.

I am pretty sure that I've had Facebook posts blocked with some sort of "Deemed false information by Facebook", which was almost certainly determined by some sort of AI.

[Coda]

For AI to work at optimal levels it's important to provide very specific criteria when asking for advice.  For example, instead of asking ChatGPT to list the best value audio equipment for sale today with a total price of $1,000 or less, ask the same question more like this: 
 

Act as an expert audio equipment advisor. Recommend 4-6 complete stereo system setups that include a dedicated CD player (or integrated CD functionality). Each setup should prioritize high-rated sound quality (based on user reviews, expert ratings from sites like What Hi-Fi?, CNET, or Audioholics, aiming for 4+ stars average), while offering excellent value for money—meaning a strong balance of performance, features, and price (target total budget per setup: $500-$1,000, avoiding ultra-premium or budget junk).For each recommendation:

• List the key components: receiver/amplifier, speakers (bookshelf or floor-standing), CD player, and any essential cables/accessories.
• Include popular brands like Yamaha, Denon, Onkyo, Sony, Cambridge Audio, or similar.
• Provide approximate current prices (based on major retailers like Amazon, Best Buy, or Crutchfield) and total estimated cost.
• Highlight pros (e.g., sound clarity, bass response, build quality) and cons (e.g., any limitations in connectivity or power).
• Explain why it offers great value (e.g., "Punches above its weight in audio fidelity for the price").
• Suggest where to buy and any current deals.

Base recommendations on 2023-2026 models with modern features like Bluetooth, streaming support, or HDMI, but ensure CD playback is central. Prioritize systems with neutral, accurate sound reproduction suitable for music listening in a medium-sized room. If needed, reference recent reviews or comparisons to justify choices. End with tips on setup and any upgrades for better value.

 

This level of specificity will help you avoid AI hallucinations and AI slop.  LLMs are getting much better, and very quickly.  Tweak this prompt to suit your music tastes and post the recommendations, I'm curious about the results.  🚀

[John L]

I think it is crazy that most AI programs will have it spit complete bullshit before ever admitting that it doesn't know the answer to any question.  

 

.