Just Have AI Do It

in The Pub8 days ago

Today, I was asked to attend a Copilot AI meeting with my boss. After we were in the meeting, we started chatting about AI in general. I mentioned concerns that when done incorrectly, it's akin to crowdsourcing information from random people via Wikipedia. Also, data privacy is nonexistent. I don't dare put proprietary information into a prompt. My boss didn't disagree one bit.

We agreed it has limited uses, such as checking for grammar, rewording sentences, etc., but no real heavy lifting. I also brought up that relying on AI to do the "heavy lifting" in our positions long-term will cause our skills to deteriorate to our own detriment.

(Created with Bing)

Make no mistake. I'm not in the least bit threatened by AI. My position isn't going anywhere. AI can't even discern that an awkward pause on a basic interview question should raise a red flag as to whether or not we decide to hire someone. It's the worst "Yes Man" you could expect. Every bad idea you ever came up with, would suddenly be affirmed.

People want to think AI will do amazing things. It very well might in limited capacity, with heavy oversight. Without guidance, AI can't even create a Mad Lib for a team building exercise without getting stuck on a theme of lighthouses and space.

AI can be great. It made the main image for this post far better than I ever could. AI can help you design a car, but by the time we've become so lazy and inept that we're considering having AI drive for us, maybe we need to either hire a chaffeur or, better yet, rethink how we use it and maybe dial it back a bit.

1000017156.gif

1000033547.jpg
Buy DUO and stake to receive Hive dividends, tip others, have your posts curated, and grow Hive!
https://tribaldex.com/trade/DUO
https://hive-engine.com/trade/DUO




Do you like stacking gold and silver?

Are you new to Hive, or know someone new to Hive, and want help and support?

Have you been on Hive for a while and want to help new Hive members? Join HOC and help us grow Hive and encourage new members!

Posted Using INLEO

Sort:  

You received an upvote of 100% from Precious the Silver Mermaid!

Please remember to contribute great content to the #SilverGoldStackers tag to create another Precious Gem.

AI can have its uses. However, like you said it has limitations and security risks as well. People need to approach its use with care, especially for and with business.

Exactly. I'll ask it to reword something with non-proprietary information or summarize a meeting, but that's about it.

!BBH
!PIZZA
!ALIVE

In ten years it won’t.

Spot on about the 'Yes Man' effect. I’ve noticed that if you feed AI a flawed premise, it’ll spend five paragraphs hallucinating reasons why you’re right instead of telling you it’s a bad idea. That lack of critical friction is dangerous for high level decision making i believe.

Yeah, it's scary. To make things worse, a lot of the people using AI don't question or check the accuracy of what they're being told.

!BBH
!PIZZA
!ALIVE

AI is like a rope with both ends on fire. I used it for a time (it was required by who I worked for) and felt like I went dumb a notch. It is great. But not something to rely on.

!LOL "Felt like I went dumb a notch" is a great explanation! I'm going to use that! 😆

!BBH
!PIZZA
!ALIVE

PIZZA!

$PIZZA slices delivered:
@bulliontools(3/5) tipped @deraaa
bulliontools tipped coinjoe
bulliontools tipped zubygeee

Please vote for pizza.witness!

Wow what a surprise, besides .80 cents all your votes are bought by u. Leo with a monthly subscription and Ed’s & SGS with delegation. You are milking rewards and posting trash 🗑️. What a joke! No wonder u make so many mistakes , ur maiming it in and just post anything to get ur paid for top votes that upvote anything.

IMG_6738.jpeg

I'd rather rely on wikipedia than any LLM marketed as AI

the way wikipedia uses a hive-mind to check the veracity of information provided has become legendary and, while mistakes do still happen, they are often rectified within a few days.

The Large Language Models touted as AI, on the other hand, have been proven to make up things and, from an experiment I conducted last year, struggle to follow explicit instructions while completing tasks.

Right? At least with Wikipedia there is a general effort to maintain factual accuracy, so there are checks and balances there.

!BBH
!PIZZA
!ALIVE