ChatGPT

[jesmu84]

Interesting article by a guy who sold his home using AI as his "realtor":

https://www.nytimes.com/2026/05/28/technology/sell-house-with-ai-no-realtor.html?

Paywall

[NCMUFan]

But was there a breathing closing attorney?

[PointWarrior]

Probably, but a lot less breathes by the attorney during this transaction.

But was there a breathing closing attorney?

[MU82]

From Axios:

AI tools might be hallucinating less. But they're still spitting out inaccurate answers cloaked in polished, hyper-confident language.

Why it matters: The more people trust AI, the less likely they are to catch costly mistakes. It's a growing problem as people increasingly lean on the technology for research, medical advice and schoolwork.

The big picture: Obvious hallucinations are easy to catch. The real trouble comes from false answers that sound convincing.

Plausible citations, mostly correct summaries, and confidently wrong answers slip past users.

If AI becomes accurate enough often enough, people might stop fact-checking altogether.

[PointWarrior]

Easy answer - actually read the AI output. Bad AI output is either lazy input, lazy prompting, or lazy checking.    But that's most of the workforce today...

From Axios:

AI tools might be hallucinating less. But they're still spitting out inaccurate answers cloaked in polished, hyper-confident language.

Why it matters: The more people trust AI, the less likely they are to catch costly mistakes. It's a growing problem as people increasingly lean on the technology for research, medical advice and schoolwork.

The big picture: Obvious hallucinations are easy to catch. The real trouble comes from false answers that sound convincing.

Plausible citations, mostly correct summaries, and confidently wrong answers slip past users.

If AI becomes accurate enough often enough, people might stop fact-checking altogether.

[NCMUFan]

Easy answer - actually read the AI output. Bad AI output is either lazy input, lazy prompting, or lazy checking.    But that's most of the workforce today...

But do you do more than read the AI output?  Do you fact check the output AI provides you?
AI has a huge ego for not being wrong and pinning it on only bad prompts seems pretty lazy in itself.

[MU82]

Fear of AI is bringing young Americans together.

Harvard's youth poll found 59% of Americans 18 to 29 see AI as a threat to their job prospects, including 66% of young Democrats and 59% of young Republicans.

[PointWarrior]

yes, read, fact check, rewrite parts every time....  it gets you 90% of the way there.  and that 90% about 100x faster...

But do you do more than read the AI output?  Do you fact check the output AI provides you?
AI has a huge ego for not being wrong and pinning it on only bad prompts seems pretty lazy in itself.

[mu_hilltopper]

So .. the other day, I read an article and 10 minutes later, had a 9 billion parameter LLM running on my laptop.  It's not the latest and greatest, but then again, nor is my humble 16GB laptop.

NVIDIA has a product called the DGX Spark, which is about the size of a dictionary, sells on Amazon for $4700 and is rated at "1 petaflop of AI performance."

These $4700 boxes can surely handle multiple AI queries at a time, let's just go with a nice round number of 10 users per box.  $500 per employee is not even a rounding error. 

Yes, the models that ChatGPT, Claude, Gemini are "better" but all models are catching up fast.

So .. why are "hyperscalers" building out dozens and dozens of new data centers across the country?

Isn't the obvious path for businesses to buy these $5k boxes and slice down their huge AI subscription fees?

[Hards Alumni]

So .. the other day, I read an article and 10 minutes later, had a 9 billion parameter LLM running on my laptop.  It's not the latest and greatest, but then again, nor is my humble 16GB laptop.

NVIDIA has a product called the DGX Spark, which is about the size of a dictionary, sells on Amazon for $4700 and is rated at "1 petaflop of AI performance."

These $4700 boxes can surely handle multiple AI queries at a time, let's just go with a nice round number of 10 users per box.  $500 per employee is not even a rounding error. 

Yes, the models that ChatGPT, Claude, Gemini are "better" but all models are catching up fast.

So .. why are "hyperscalers" building out dozens and dozens of new data centers across the country?

Isn't the obvious path for businesses to buy these $5k boxes and slice down their huge AI subscription fees?

That depends.  The federal government as your client ensures a constant stream of cash flow.  Kind of like RTX, Boeing, NG, etc...

Should these companies have something like what you're describing as well?  Yes, I think so.

I think the real problem with OpenAI and Anthropic, etc is their models aren't getting better.  And they aren't making money.  That's why they want to IPO ASAP.

[jesmu84]

That depends.  The federal government as your client ensures a constant stream of cash flow.  Kind of like RTX, Boeing, NG, etc...

Should these companies have something like what you're describing as well?  Yes, I think so.

I think the real problem with OpenAI and Anthropic, etc is their models aren't getting better.  And they aren't making money.  That's why they want to IPO ASAP.

So someone else can be left holding the bag?

[Hards Alumni]

https://thezvi.substack.com/p/white-house-will-ad-hoc-decide-who?ref=pdoom100.com

[MU82]

https://thezvi.substack.com/p/white-house-will-ad-hoc-decide-who?ref=pdoom100.com

Socialism.

[NCMUFan]

So .. the other day, I read an article and 10 minutes later, had a 9 billion parameter LLM running on my laptop.  It's not the latest and greatest, but then again, nor is my humble 16GB laptop.

NVIDIA has a product called the DGX Spark, which is about the size of a dictionary, sells on Amazon for $4700 and is rated at "1 petaflop of AI performance."

These $4700 boxes can surely handle multiple AI queries at a time, let's just go with a nice round number of 10 users per box.  $500 per employee is not even a rounding error. 

Yes, the models that ChatGPT, Claude, Gemini are "better" but all models are catching up fast.

So .. why are "hyperscalers" building out dozens and dozens of new data centers across the country?

Isn't the obvious path for businesses to buy these $5k boxes and slice down their huge AI subscription fees?

https://www.youtube.com/watch?v=aXy8mQeuObk

[mu_hilltopper]

https://www.youtube.com/watch?v=aXy8mQeuObk

Lol .. that exact video was what inspired me to install the LLM on my laptop.