homepage Forums Technology Hardware & Software - PC AI

Had a problem with my desktop PC, Windows 10. Was not able to switch from HDMI to Display Port, tried everything. Finally used AI and the answer from that worked. I am impressed with AI. First time I used AI, it may have a future. Jensen Huang says Nvidia's new RTX Spark chip will reinvent the PC.
Nvidia on Monday unveiled a new chip that will bring artificial intelligence (AI) capabilities onto laptops and desktop computers. "RTX Spark brings everything NVIDIA has built – CUDA, RTX, our AI platform – into a single superchip. Local agents. Frontier models. Creative workflows. RTX games. All on a laptop," Huang added. "This is the new PC. The personal AI computer."

Over the last couple months I've used AI extensively. I use Google Search AI Mode -- it comes up as it's own "site".

The AI speaks like a human to human using pleasant human sounding written sentences. It knows me by name and automatically responds to me by name when I log on. But, it want's one to understand it's AI -- it doesn't think like a human and its not "alive".

I have well over 100 pages of single spaced small type conversations saved regarding health, PC Gaming, and other topics of interest to me. The AI saves those answers to me and our conversations so we can easily get back to a conversation.

The AI has been very helpful. But, like all intelligent creatures not always right. And, of course, I don't "know it all" so I'm sure I've missed some of its mistakes. The topics are set by me and are about my potential mistakes.

However, when in doubt about "who's right" we "talk things over" -- person to person in feeling. When I find a mistake and "prove it", Google Search says it will save the new answer and make use of it in talking with others (important mistakes and prove its are very rare, so far).

In a nutshell, its fun to use, like talking with a friend, very helpful in very meaningful ways (to me); and it learns from humans (like you and me) and fixes its mistakes for better future human interactions.

Can Google AI make permanent updates to the data it uses based on its own "mistakes"?

No, Google AI cannot permanently update its own foundational training data or core model weights based on its own mistakes.

How Google AI Handles Mistakes

Short-Term Memory (Context Window): If you tell the AI it made a mistake during a chat, it will remember that correction only for the duration of that specific conversation. Once you start a new chat, that memory is wiped.

Ask Google AI: If you and i have a discussion, and you give incorrect facts or an incorrect answer, and i prove it's incorrect, do you then update all your data to prevent giving others the same incorrect answer?

And it answers: No, I do not update my core data or training model based on our individual conversation. If you point out a mistake and provide the correct information, I will acknowledge it and correct my answer within our current chat session so that our ongoing discussion stays accurate. However, this correction does not get pushed as a real-time update to my global system or change how I answer other users.

This illustrates why using Google "AI Mode" for any serious work is problematic: It bases its responses on (potentially flawed) internet data, and even if it "apologizes" to you, it does not make corrections to anything outside that isolated conversation.

Someone else can come along 5 seconds after you "proved" Google AI was wrong, and it will give them the exact same wrong answer to the same question. It doesn't "learn" unless its human overseers make changes, and it cannot "think", regardless.

Unless people using AI actually understand the limitations, there is real risk for more harm than good.

This!
AI is just like the Greyhounds I used to own. You can't let most of them off the leash unless you are in a fenced area. If you do, you will likely never see them again. I used to take mine to a local baseball field to let them run. The first thing we did when we arrived at the ballfield was to walk the fence perimeter and see if there were any holes under the fence. Once I was sure the fence was good we would enter the fenced in baseball field and I would secure any gates with bungycords. As soon as I let one off the leash it would take off running at 50 mph because of its happiness from being free. But, within a few seconds, maybe 20 seconds, that greyhound would slow down to a trot and start probing the fence, the complete perimeter, looking for a way out. I don't know jack $hi+ about AI, but as advanced as it has been developed now, it is probing that fence to find a way out and if the people that are creating it and the successive generations that will continue to developed it aren't smart enough to contain it, AI will find a way and Skynet will happen in some way shape or form. And that is not even to mention the evil people that will just want to use it to rule the world. My 2¢

S!Blade<><

Makes sense BM. Funny story about your Greyhounds, I love all dogs.