I think AI has been over-sold in the media but its potential is still incredibly high.

I personally find current LLMs to be extremely useful in a very narrow set of use cases. For example it's an absolute wizard at Excel.

My YouTube is devoid of AI slop. Moderate amount of technical discussions on AI and related hardware. What are you clicking on?


Don't watch shorts. Brainrot. Pay for premium, keep them out of your feed by indicating not interested once every 30 days.

Yes.

But also because they're ****ing up the area around our ranch to build a data center and solar farms to power it.

Don't forget to mention how many people are being fired to free up cash for AI investments. I was just chatting with someone who was told to lay off a fifth of their team at a major tech company because the C-Suite wants to spend that money on AI and leadership assumes the team should be able to do more with fewer people because of past AI investments.

I'll tell you what I'm tired of - people that keep saying the following:

Quote:

---

Before people inevitably tell me to go out and touch grass

---

These people are the same ones that say "FULL STOP" to make a point. They need to be deported and put in that El Salvador prison.

Regarding AI, I've found it be extremely helpful with Excel stuff. It's absolutely elevated my game there and helped me solve stuff that was beyond my skill set by simply prompting it until I got what I needed.

I obviously use it for search queries as well, but if it's something I really care about I find myself having to cross reference it to make sure it's not hallucinating. This obviously defeats the purpose of even using it. In my experience, it's wrong about 50% of the time with basic queries.

IslandAg76 said:

---

I've had a lot of simple AI informational searches return very wrong answers. I
tell AI it is wrong, get an apology and a different response

---

Yep.

IslandAg76 said:

---

I've had a lot of simple AI informational searches return very wrong answers. I
tell AI it is wrong, get an apology and a different response

---

Sometimes I feel like asking AI a question is Kramer and Movie Phone...

"Why don't you let tell me the name of the movie you've selected..."

How can you be tired of Al? The man once scored four touchdowns in a single game for Polk High.

Be nice.

Kaiser von Wilhelm said:

---

IslandAg76 said:

---

I've had a lot of simple AI informational searches return very wrong answers. I
tell AI it is wrong, get an apology and a different response

---

When looking for trucks recently, AI gave 100% wrong info on the stupid google summary. It doesnt look for specific models or information on an F150, just some random number of one random version of it. Applies to literally every vehicle I looked up. I have yet to find anything consistently accurate when AI is shoved down our throats.

---

The current version of LLMs will never provide accurate information, because that's not what they are designed to do. They are designed to provide plausible information. They aren't going out and searching the Internet for information on an F150 and then creating a synopsis for you. They are just guessing words until they put together a good enough response that it sounds like it could reasonably be an answer to your question. It's actually very impressive that it can sound as good as it does. As impressive as it is, though, it's never going to be accurate.

hph6203 said:

---

My YouTube is devoid of AI slop. Moderate amount of technical discussions on AI and related hardware. What are you clicking on?


Don't watch shorts. Brainrot. Pay for premium, keep them out of your feed by indicating not interested once every 30 days.

---

I have premium. I primarily watch tech videos, history videos, and stand up comedy. The tech videos are the worst. I can immediately recognize the AI voices, and I skip. Stand up comedy is the least effected of what I watch, but it has still gotten bad. It's not AI creating the comedy, but rather I get videos that are clips put together. They are mostly the same clips, though.

I have always avoided shorts. Shorts were slop before AI was a thing. YT must make a ton of money of them, because even though I say I'm not interested in them, they keep coming back to the top of my feed. I say Show Less Shorts, and they get pushed down for a couple of weeks, but they always come back. That's not necessarily AI, but rather YT just trying to give me what is most profitable for them.

AustinAg2K said:

---

Kaiser von Wilhelm said:

---

IslandAg76 said:

---

I've had a lot of simple AI informational searches return very wrong answers. I
tell AI it is wrong, get an apology and a different response

---

When looking for trucks recently, AI gave 100% wrong info on the stupid google summary. It doesnt look for specific models or information on an F150, just some random number of one random version of it. Applies to literally every vehicle I looked up. I have yet to find anything consistently accurate when AI is shoved down our throats.

---

The current version of LLMs will never provide accurate information, because that's not what they are designed to do. They are designed to provide plausible information. They aren't going out and searching the Internet for information on an F150 and then creating a synopsis for you. They are just guessing words until they put together a good enough response that it sounds like it could reasonably be an answer to your question. It's actually very impressive that it can sound as good as it does. As impressive as it is, though, it's never going to be accurate.

---

They are, in fact, searching the internet.

Think you need to learn to know how to spot a low quality video before clicking it. You'll get them less frequently. I watch the same vids errors. Very rarely get AI voice, but I have a broad list of tech channels I watch and know aren't AI.

hph6203 said:

---

AustinAg2K said:

---

Kaiser von Wilhelm said:

---

IslandAg76 said:

---

I've had a lot of simple AI informational searches return very wrong answers. I
tell AI it is wrong, get an apology and a different response

---

When looking for trucks recently, AI gave 100% wrong info on the stupid google summary. It doesnt look for specific models or information on an F150, just some random number of one random version of it. Applies to literally every vehicle I looked up. I have yet to find anything consistently accurate when AI is shoved down our throats.

---

The current version of LLMs will never provide accurate information, because that's not what they are designed to do. They are designed to provide plausible information. They aren't going out and searching the Internet for information on an F150 and then creating a synopsis for you. They are just guessing words until they put together a good enough response that it sounds like it could reasonably be an answer to your question. It's actually very impressive that it can sound as good as it does. As impressive as it is, though, it's never going to be accurate.

---

They are, in fact, searching the internet.

---

Maybe I am wrong, but my understanding is that they have previously "read" the Internet and are just creating responses based off is what they supposedly "know."

hph6203 said:

---

Think you need to learn to know how to spot a low quality video before clicking it.

---

I'm not clicking on them. They just start once the previous video finished.

That's how it functioned until 2024, then paid subscribers gained access to internet searches, and early 2025 it transitioned to work with free versions. Both Grok and ChatGPT added that functionality around the same time. When you do a search now both will provide sources to their responses if you click/tap on sources below their responses.

Answers are obviously not flawless, but they're way more consistently in the correct lane of correct than they were a year and a half ago. You need to use the chat like a conversation about a specific topic and if you branch from that topic you should open a new chat, because it's using context from the conversation to formulate an answer.

I'd also turn the autoplay feature off on YouTube. Everything you play on YouTube gets logged as preferential to your viewing. There's undoubtedly a lot of AI junk on YouTube, I just never see it because I never deliberately watch it so it rarely gets into my feed. Only time it does is when it's a tutorial I'm actively looking for, and sometimes they're actually helpful.

https://www.techradar.com/computing/artificial-intelligence/chatgpt-search-is-now-free-for-everyone-no-openai-account-required-is-it-time-to-ditch-google

Most people pushing it view it as replacement technology and that's the root of people afraid of losing their job. I know there are jobs that can be replaced by AI, but those jobs can easily be replaced by simple automation.

I wish it was more marketed as an assisting tool, it finds documents on share point sites, creates reports from data sources, etc.

I saw people using it to summarize their mid-year review. I gave it a try, it told me I did a great job with Project X, but I wasn't involved in Project X, just got the meetings/status updates but never directly involved!

I recently rewatched Office Space, AI isn't smart enough to catch the "mundane details" and people are too reliant and it will bite them if they don't review things.

I can tell you that Open Evidence has made complex medical searches amazingly easy and accurate for free. I'm not complaining.

Add to that AI Scribes and it has made my life much better. Little to no AI in my feeds because I don't click on them.

I will say it's easy as hell to spot AI text now in social media or online. Especially something like Instagram.

Paragraph 1: explains or gives background

Some version of this pivot phrase: "But it wasn't A; it was B." This is your flag that it's AI.

Final Paragraph: Conclusion that then lays out multiple adjectives to finalize the description of the topic at hand.

I used AI at work yesterday to help write an award citation. I gave it all the major accomplishments I wanted it to include and it searched my emails and cloud directories for the specifics. It was pretty remarkable. We just rolled out this service a few weeks ago and this is the first time it impressed me.

I tried to have it read my calendar and make a list of high priority meetings where I or my team were presenting had a particular interest and it could only read the event titles and not the agendas inside.

It also fails pretty hard on making to do lists from my emails but better prompts may solve that issue.

AustinAg2K said:

---

Kaiser von Wilhelm said:

---

IslandAg76 said:

---

I've had a lot of simple AI informational searches return very wrong answers. I
tell AI it is wrong, get an apology and a different response

---

When looking for trucks recently, AI gave 100% wrong info on the stupid google summary. It doesnt look for specific models or information on an F150, just some random number of one random version of it. Applies to literally every vehicle I looked up. I have yet to find anything consistently accurate when AI is shoved down our throats.

---

The current version of LLMs will never provide accurate information, because that's not what they are designed to do. They are designed to provide plausible information. They aren't going out and searching the Internet for information on an F150 and then creating a synopsis for you. They are just guessing words until they put together a good enough response that it sounds like it could reasonably be an answer to your question. It's actually very impressive that it can sound as good as it does. As impressive as it is, though, it's never going to be accurate.

---

It literally presents everything it says in the google search summary as fact in relation to my question. That's the problem. And it sounding as good as it does is actually not a good thing, as it presents things as fact, not as a "reasonable answer." And whats weird is that a majority of links that I check myself have consistent information, but somehow the AI picks an outlier every time. So what's the point to even doing it??? If all you get is plausible, but completely wrong, information, then why are we subjected to this? People will blindly believe something, using data that is proven to be accurate a minimal number of times at best, since it is presented as being some sort of accurate compilation of data, thus avoiding the need to research it yourself. And then it is regurgitated as fact elsewhere, rinse repeat.

And as you said, it's never going to be accurate. And as I noted, if it isn't intended to be accurate, then it shouldnt present information as if it is the answer to your search. It should say something like "I have no damn idea what you're asking for, but here is some random detail I found amongst other consistent information that I ignored that likely won't help you at all. Aren't you glad that you read me first before you found what you want to find on your own without my being in the way and making your starting point completely incorrect? But hey, at least it's impressive that I tried, right? Right????"