Will AI take your job?

https://www.boards.ie/discussion/comment/123982473#Comment_123982473

"Tbh I'm impressed it found anything on a German120 yr vintage lathe where most of the information is in vintage books not likely to be on the web or digitised. "

That's the problem with AI. It didn't find the information to answer the question. But instead of coming back with that, it just made up an answer. You might be impressed but i am not.

But seriously, the problem is not going to be misinformation on lathes or motorcycles, important to some of us as they are, but rather when governments and public sectors start using it. What if you are refused social welfare or a pension because of some AI bot. Or your child is not admitted to a local school because of one of these AI engines. I m banned certainly would like to see them banned from any decision making role, even if it is only a supporting one. I have no problem with AI writing code. I have seen my fair share or rubbish code written by humans and I imagine AI could match it. I would also like to see consequences for misinformation so that if an answer was wrong the AI engine owners would be liable for any damage caused. Of course that won't happen, IT companies seem to have governments around their little fingers and are generally excused their liabilities, for example on online publishing.

PS. Your links, Vic is an old friend and i could of course have asked the question of him .

https://www.boards.ie/discussion/comment/123982315#Comment_123982315

I agree it's a poor example. But it seems like a good for fit travel booking systems in general. All data heavy.

https://www.boards.ie/discussion/comment/123982311#Comment_123982311

https://sites.google.com/view/pittler-lathes/accessories-dropdown-menus/gears-worms

https://sites.google.com/view/pittler-lathes/accessories-dropdown-menus/pittler-literature

"...There is not a great deal of vintage Pittler literature easily available in English unfortunately. ..."

Tbh I'm impressed it found anything on a German120 yr vintage lathe where most of the information is in vintage books not likely to be on the web or digitised.

And it's an unconventional lathe to begin with. It's hardly "my first lathe". I've used conventional lathes not a lot but I have used them in machine shops when prototyping. But decades ago. So if theres nuances about a rare vintage lathe I'm hardly going to know about it.

plodder

I heard someone talking about "Agentic AI" on Newstalk this morning. I really struggle to see what he was describing has to do with AI at all. He gave the example of querying (some unidentified AI bot) how much paid holidays you have left, and then how an AI agent might take it a step further, by both booking your time off and booking a holiday for you. I think it was a poor example, because who would actually do that? Maybe it's a good illustration of the gap between the promise of AI and the reality. I'm sure an AI agent could be made to book you a holiday (instead of the time consuming manual research that usually goes into it). But what would you end up with? Some weird mystery tour …

And what has any of that got to do with data models and so on?

https://www.boards.ie/discussion/comment/123982268#Comment_123982268

May be my original post wasn't clear. I did specify the make of motorcycle in my first query and the make of lathe in my second. I did not expect AI to read my mind.

Yes, getting the wrong information from a book can be dangerous. For one motorcycle I am aware of, the manufacturer provided a wrong wheel assembly diagram. But there were multiple corrections published subsequently which is what you expect with old-fashioned manuals. But I am not too sure how that will work with AI.

Here is the question "how do i cut metric threads on a Pittler B2 lathe?" about metric threads asked from Copilot in windows. I originally asked the question of Google on a Linux machine

To cut metric threads on a Pittler B2 lathe, you need to configure the gear train for metric pitch, use a 60° threading tool, and engage the leadscrew properly—often without using the thread dial.

Here's a step-by-step guide tailored to the Pittler B2 lathe and general metric threading principles:

🛠️ Setup for Metric Threading on a Pittler B2

Gear Train Configuration:
- The Pittler B2 lathe typically uses change gears to set the correct pitch.
- You’ll need a gear chart (often found in the lathe’s manual or on a plate) to determine the correct gear combination for your desired metric pitch.
- If your lathe has an imperial leadscrew, cutting metric threads requires a 127-tooth conversion gear to compensate for the inch-to-mm difference.
Tooling:
- Use a single-point threading tool with a 60° included angle, which matches the ISO metric thread profile.
- Ensure the tool is ground properly and set at center height.
Lathe Settings:
- Set the compound rest at 29° to allow for gradual cutting on one flank.
- Set the carriage feed to manual and engage the half-nut lever to synchronize with the leadscrew.
Thread Dial Consideration:
- If your Pittler B2 has a metric leadscrew, you can use the thread dial.
- If it has an imperial leadscrew, you must leave the half-nut engaged and reverse the spindle between passes to maintain synchronization.

⚙️ Cutting Process

Mark and Measure: Lay out the thread length and verify the pitch.
Align Tool: Position the threading tool at the start point.
Engage Half-Nut: Start the lathe and engage the half-nut at the correct point.
Make Passes: Cut in multiple passes, increasing depth slightly each time.
Reverse and Repeat: If not using a thread dial, reverse the lathe to return to the start without disengaging the half-nut.
Check Fit: Use thread gauges or mating parts to verify fit.

Which, I am sure you will agree, is a totally rubbish answer. But it would be misleading to anyone who, unlike you and I , doesn't actually know the Pittler Lathe and how to operate it.

Not entirely sure how you expect an answer to have information on a specific engine when you don't say which it is. Or specific issue when you don't mention that either. Neither AI or anyone else can read your mind.

Spin? I simply test it and see for myself what it can and can't do. Then tweak it until it works.

You could read the wrong book or look up the internet and get bad information. Does that mean the book is wrong or dangerous. Of course not.

https://www.boards.ie/discussion/comment/123982178#Comment_123982178

And that underlines the problem wit AI. Rather than suggest it doesn't know, or needs more information, it spews out irrelevant or wrong information and then considers the question answered.

It's called AI hallucination and part of the reason why any answer you get from an AI just cannot be trusted. The tendency for it to fill in gaps with reasonably sounding nonsense or, worse still, just make up stuff wholesale is a very serious problem.

https://www.boards.ie/discussion/comment/123982025#Comment_123982025

No , i don't need any help. Your post is both wrong and unhelpful for the engine I asked about. And the information was totally wrong for the lathe.

That you could find a starting procedure for a Vincent engine in a BSA frame ( just how odd is that, how many A10's are fitted with a Vincent twin engine ?) does nothing to help me with my engine.

And that underlines the problem wit AI. Rather than suggest it doesn't know, or needs more information, it spews out irrelevant or wrong information and then considers the question answered. The answer you post would be only useful to someone who actually knows how to start the engine and recognise what AI is getting wrong. Assuming i asked about something i knew nothing about, I could have absolutely no confidence that the answer from AI was in any way accurate or relevant to my question. If you cannot see that such misinformation is wrong or dangerous than I think you have obviously accepted the AI spin.

That's nice…

"...To start a BSA motorbike fitted with a Vincent Black Shadow engine, follow a careful kickstart procedure that includes priming, ignition setup, and throttle control.

Here’s a detailed breakdown of the startup process for this hybrid classic—typically a Vincent Black Shadow engine mounted in a BSA A10 frame:

🏍️ Starting Instructions for a BSA-Vincent Black Shadow Hybrid

Preparation

Ensure fuel is on: Open the fuel taps on the tank.

Check oil levels: Vital for older engines.

Ignition off: Start with the ignition switch turned off.

Priming the Carburetors

Use the ticklers on the Amal carburetors to flood them slightly.

This ensures the engine has enough fuel to fire on the first few kicks.

Compression and Kickstart Setup

Gently kick the starter to find top dead center (TDC).

Use the decompression lever to ease the piston just past TDC.

This sets up the engine for a full kick with maximum force.

Ignition On

Turn on the ignition once the engine is primed and positioned.

Make sure the throttle is slightly open—not too much.

Kickstart

Deliver a strong, smooth kick through the starter lever.

The engine should fire within a few kicks if properly primed.

Warm-Up

Let the engine idle and warm up gradually.

Avoid revving hard until oil circulation is confirmed and the engine reaches operating temperature.

This startup method is demonstrated in this video tutorial featuring a 1952 Vincent Black Shadow Series C engine. Another example of this setup in action—an 1110cc Vincent engine in a 1962 BSA A10 frame—is shown here.

Would you like help finding a workshop manual or tuning guide for this hybrid setup?..."

https://www.boards.ie/discussion/comment/123969747#Comment_123969747

Reading this thread, I decided to ask Google AI a couple of questions. The first was how to start a particular make of vintage motorcycle. The second was how to cut metric threads on a specific make of old lathe. Both answers were nonsense.

I have no doubt that AI might be useful for certain matters but not for everything. By the way, both questions would be considered legitimate questions a new owner might ask.

SpoonyMcSpoon

The fact that all these US AI companies are run by Israeli-US dual citizens is another red flag since the tide is going out on Israel’s influence on the US; no longer is it under the radar how much Israel controls the US and of course more and more people do not trust Israel. When we see how much AI is being pushed and look at who is actually benefitting from it, alarm bells should ring.

The energy conundrum is still the biggest concern however and until someone solves this, the AI bubble will burst by 2027, taking the US stock markets with it.

rogber

Lot of tech stocks sold off this week. Interesting if the bubble is about to burst or if it's just a blip. Everyday there's an article about AI revolution about to change everything and another article saying how poor AI still is at so many tasks and the LLMs are reaching their limits with disappointing results.

You really wouldn't know what to believe

Oscar_Madison

https://www.boards.ie/discussion/comment/123970417#Comment_123970417

Checking shelves is a laborious task - whilst yes, the company saves some money on staff costs, they hand it back out paying for these robots and whatever service package comes with them- so potentially some better jobs created so not all bad.
It will be years before supermarkets tackle re-stacking shelves using robots- they’ll likely need bigger premises and to design the shop around that - so that will still be a human job.

I see that shelf checker as progress- humans are made for better things

Emblematic

https://www.boards.ie/discussion/comment/123970417#Comment_123970417

Supervalu near me (a fairly small one) only seems to have whole bean coffee, instant coffee and those pods. It doesn't have ground coffee for some reason. Maybe the robot will sort this out.

https://www.thejournal.ie/supervalu-robots-ai-tally-6710762-May2025/

Robots.. coming to a shop near you...

https://www.boards.ie/discussion/comment/123970315#Comment_123970315

I don't feel particularly bent toward putting any more time or effort into AI Frank. I'm not invested and there's plenty of places to get information from which are far more reliable too. What's remarkable, though, is someone getting their knickers in a twist because someone else said that they wouldn't blindly trust what an AI said to them.

That's just weird.

AI will change things. I think everyone is in agreement there. Whether that's for good or not in the grand scheme of things remains to be seen however.

This is an interesting read, and pretty much how I'd see things.

https://simonbatchelar.co.uk/why-does-ai-get-things-wrong/

The ending of the article in particular is in line with my thinking on using AI. "Double check the facts" should be the mantra for anyone asking AI anything.

Fair enough I apologize if thats offensive. I'll quit this line of conversation.

https://www.boards.ie/discussion/comment/123970335#Comment_123970335

And now you're acting like a twat.

https://www.boards.ie/discussion/comment/123970288#Comment_123970288

Tony basically wrote off AI as useless

No he didn't.

https://www.boards.ie/discussion/comment/123970355#Comment_123970355

The AI investment bubble is whole different thing. The technology will improve incrementally regardless. The military is investing in it also. Remember the .com bubble. Things still processed.

silliussoddius

https://www.boards.ie/discussion/comment/123969747#Comment_123969747

Stop what? The current AI companies haven't got the cash for their proposed capital expenditure.

https://www.boards.ie/discussion/comment/123970315#Comment_123970315

Tony's been through a few world wars he's robust. He's a regular on the mil forums. He'll know about AI in the military.

https://www.boards.ie/discussion/comment/123969548#Comment_123969548

Even in Copilot switching to GTP-5 is another leap. Or using Copilot Studio to tweak your agent vs the stock agents. In the last 3-4 months it's taken leaps forward. It can do all sorts of office or coding tasks it couldn't to 4-6 months ago.

Is it perfect no. Is it useful most definitely. Even Gemini has got a lot better. But do I fact check Gemini. yes I do.

Frank Bullitt

https://www.boards.ie/discussion/comment/123970288#Comment_123970288

Fair enough. He might well put more time and effort into it, and it might work better. Still no need for the holier than thou firebacks to him though.

I agree AI will change things, it already has. I do not agree with the dystopian preaching that it will take over the world.

https://www.boards.ie/discussion/comment/123969961#Comment_123969961

That's disingenuous.

Tony basically wrote off AI as useless he barely scratched the surface on a lightweight version of it a while ago (now obsolete), and that was pointed out to him. That if he invested more effort into it, and a current or different version, he would have better experience of it.

And I'm not the only one saying the same thing. No one's using any buzzwords. Dodgy analogies for sure.

Tony might know a P-51 from a A-36 but he's mistaken about AI.

Frank Bullitt

https://www.boards.ie/discussion/comment/123969771#Comment_123969771

I get a laugh out of posters like this.

“If only you knew what I know…”, and then just post buzzwords and short responses to give off a whiff of knowledge. Pretty amusing stuff.

You're eating McDonald's thinking it's food.

https://www.boards.ie/discussion/comment/123969747#Comment_123969747

Don't try and tell me what I've seen or haven't seen.

I've seen AI completely make stuff up and present it as fact.

https://www.boards.ie/discussion/comment/123969525#Comment_123969525

You've seen nothing. Sticking your head in the sand isn't going to stop it.