Why AI still has a long way to go (and you shouldn't trust it much)

[sgt_zim]

I can't find the thread about AI, maybe it was in politics. Not going to search that thread.

Anyway, I asserted then it's little better than Encyclopedia Brittanica or World Book, and that's still the case.

From Grok (X's AI engine) on 7mm RM vs 7mm PRC

The 7mm PRC, introduced by Hornady in 2022, and the 7mm Remington Magnum, launched by Remington in 1962, differ significantly in design and application despite their shared 7mm caliber.

The 7mm PRC features a modern, non-belted case designed for short-action rifles, making it more compact and lighter, while the 7mm Remington Magnum uses a belted, long-action case, resulting in heavier, longer rifles.

The PRC’s case holds about 77 grains of powder, producing velocities around 2,950-3,000 fps with 175-grain bullets, whereas the Remington Magnum’s larger 82-grain capacity pushes similar bullets at 3,000-3,100 fps, but with higher recoil.

The PRC is optimized for precision with high-BC bullets, offering flatter trajectories and longer barrel life due to its efficient design, while the Remington Magnum, though accurate, is less tailored for modern precision shooting and has a shorter barrel life due to higher pressures.

Ammunition for the PRC, being newer, is less widely available and often premium-priced, whereas the Remington Magnum benefits from decades of widespread, diverse factory loads.

The PRC’s SAAMI specs cater to modern CIP-length magazines, ideal for long, heavy bullets, while the older Remington Magnum can be less compatible with such projectiles in some rifles.

Ultimately, the 7mm PRC excels in long-range precision and modern rifle systems, while the 7mm Remington Magnum remains a versatile, general-purpose magnum for hunters prioritizing availability and proven performance.

Knowing nothing about 7mm PRC, but plenty about 7mm RM, and also understanding the Gell-Mann amnesia effect, I have to assume much of the info on 7mm PRC is as wrong as it is on 7mm RM.

 

[GuttormG]

Tested AI quite extensive at work, it is 99% waste of time.
For those in need of a bullshitgenerator it is a hit!

 

[analog_peninsula]

AI / machine learning is an amazing tool, but only a tool, and is subject to the limitations of any other tool in that it’s best at solving certain types of problems. Most of the public discussion feels caught up in the Gartner Hype Cycle peak of inflated expectations, but it’s worth remembering that using the various AI agents for tasks at which they aren’t really suited is like driving screws with a hammer or pounding nails with a screwdriver. Most of the press treats the technology like it’s a form of magic because they’re largely technical illiterates.

 

[Altitude sickness]

I hope it’s not watching this thread.

 

[BRICKBURN]

Something beyond the standard LLM and also "gives credit" and attribution to the actual source of the information being provided.
https://www.perplexity.ai/

 

[wesheltonj]

I've been to several classes on AI. It just makes stuff up that's not correct. Can alter photos (that look real) and video (not quite real looking yet, but getting there). It going to be a real problem for my business. Thankfully, it's not my responsibility to root out the fakes just "punish" the folks who submit the fakes.

 

[LimbNoMo]

My nephew recently planted some roses in his backyard. He took photos with his iPhone. It distinctly added blossoms that weren’t there. Both the color and petal shape were wrong.

 

[analog_peninsula]

Again, the right tool for the right job. Generalizing in this case isn’t terribly helpful. Go to 1 hour 6 minutes in the following video for a very eye opening demo. Figure six or seven minutes to watch.

 

[Hogpatrol]

Have a bud that has both the 7RM and the 7PRC. The OP seems to be mostly correct.