Let's talk about Artificial Intelligence

[Jerry_Atrick]

There is an article in The Age on the subject matter today: https://www.theage.com.au/technology/kids-playing-with-a-nuclear-warhead-why-we-must-heed-ai-warnings-20260910-p60w2j.html

 

I think the extinction thing is a little overkill (pardon the pun) unless robotics makes incredible strides over a short period of time. However, it won't be as Holywooders and Youtubers would suggest that roboits go rogue or swarms of drones. It will be their ability to crack the systems; of course, there would be real problems if they crack military systems and start guiding nuclear weapons, for example. 

 

But, if the world is going to end in 3.5 years for us, may as well take out a 10 year loan on that Ferrari, with a deferred payment for 5 years and live a little 😉

 

[pmccarthy]

I think any car pre about 1984 is ok as they have no computer on board. My house is a greater threat as my vacuum cleaner and pool robot could rise up against us.

[rgmwa]

So you are hiring 100 AI agents disguised as humans? 

[Jerry_Atrick]

I wish.. AI agents aren't (outwardly) so temperamental.. and they get the issues. 

 

 

[rgmwa]

What are you doing that needs 100 new hires? 

[Jerry_Atrick]

Developing and implementing all risk systems for our whole European business.

 

Edited 21 hours ago by Jerry_Atrick

[nomadpete]

Getting back to the cost/benefit equation. Exactly how is A.I. and  data centres going to make life better for the common man? (i.e, the majority of people).

[Jerry_Atrick]

That does not depend on AI.. that depends on those who govern us. And history shows that pollies wait until the problems are well entrenched before they deal with them. 

 

But also, even if the governments finally swallowed the need to tax the corporates to fund more and more people with nothing to do.. what is the impact on those people effectively on the scrap heap. Yep, some enterprising souls will thrive, but we are mainly conditioned to go to school, get a job, work and then retire. So, the way we educate future generations will have to change, too. 

 

A friend of mine in a more senior position at  another bank is telling me they have gone all in on AI for software development and it is giving them between 15 and 20 x the productivity. So he will end up implementing a decent swathe of cuts but hold enough to significantly improve rate of output. 

 

I have been saying for a couple of years that in the next now 8 - 13 years, there won't be many software programming languages left. People will type in their prompts and the run time will be generated directly from it. Yes, some of the lower level languages will still be around, but there will come a time where they are no longer needed as well. 

 

Certainly in software development - ironically the industry that gave birth to AI - there isn't much for the common software developer - they will need to be English (or other human language) graduates

 

[nomadpete]

In summary, you haven't listed any personal benefits to the masses.

 

But you do hint that big business such as banking will make more profit than ever, and employ less humans than ever.

 

Yet we are constantly told how great things will be after this revolution

[old man emu]

12 hours ago, rgmwa said:

What are you doing that needs 100 new hires? 

Simply complying with  Parkinson's Laws: 

 

Parkinson's Law is the title of a satirical essay by the naval historian C. Northcote Parkinson which was published in The Economist in 1955, following the report of the Royal Commission on the Civil Service. He proposed the following Laws: 

1. "The Law of Multiplication of Work" (Work expands so as to fill the time available for its completion).

2. "The Law of Multiplication of Subordinates" (The number of workers within administration tends to grow, regardless of the amount of work to be done).

 

After observing that the promotion of employees necessitated the hiring of subordinates, and that the time used to answer minutes requires more work, Parkinson stated: "In any public administrative department not actually at war the staff increase may be expected to follow this formula" :

where:

x – number of new employees to be hired annually

k – number of employees who want to be promoted by hiring new employees

m – number of working hours per person for the preparation of internal memoranda (micropolitics)

P – difference: age at hiring − age at retirement

n – number of administrative files actually completed.

 

Here's a video on this:

 

 

[nomadpete]

Gee thanks for clearing that up, OME.

 

So, maths holds an important place in every administration.

[old man emu]

25 minutes ago, nomadpete said:

Gee thanks for clearing that up, OME.

Don't forget that the book, Parkinson's Law, was a piece of satire. But satire is not comedy. Satire's main purpose is constructive social criticism. It does not just poke fun for the sake of entertainment; it holds up mirror images of flaws in individuals, governments, or institutions to encourage reflection or change.

[nomadpete]

A long time ago, I subscribed to several 'Business Management' periodicals. Although I did this primarily to upset my paranoid manager, there were numerous interesting articles. For instance, There are long mathematical formulae for the prediction of absenteeism by workers. Much like your Parkinson's Law one.

 

Maths indeed are used by big business for some curious reasons.

 

BTW, my manager never subscribed to such magazines. He simply 'knew' everyone was after his job. Character flaws in my workplace were a constant source of amusement. Most of us were not harmed by this.

[old man emu]

7 minutes ago, nomadpete said:

Maths indeed are used by big business for some curious reasons.

I don't think Parkinson's formula was meant to be applied in the real world. It is part of the satirical story. If you think about it, many activities or situations can be boiled down to some sort of equation connecting variable factors in order to provide a prediction for a possible outcome. A common situation in basic problem solving in maths is how to calculate how long three people, working at different rates can complete a job. The job is often something simple, like mowing an area of lawn. The idea of the problem is to teach the student how to generate an algebraic equation from data provided. 

 

I find that sort of task more useful than solving an equation for the values of 'x' given the equation mx + b = 0, and the values of m and b are given.

[nomadpete]

1 hour ago, old man emu said:

But satire is not comedy

I understand Parkinson's satire. I pointed out there is a real process that is not satire. My point is that in real business, real people are doing it without seeing the silly side.

 

To me, real life contains a lot of laughable events. Real life is my comedy. For instance, most politics cannot be taken too seriously, and is best laughed at.

[willedoo]

All the talk in the news lately about artificial superintelligence and the urgent need to regulate it reminds me of a virus that's already escaped the laboratory. I get the feeling that the genie will be well and truly out of the bottle before the world's lawmakers can do anything about it.

[willedoo]

On the positive side of AI, the advances in graphics are leaping ahead.  It won't be too long before we can watch movies that don't look like a video game. It will soon get to the stage where the realism will surpass that.

[red750]

What staggers me is the way you can take a creased old black and white photo taken 70 odd years ago and carried around in someone's wallet, and in a few seconds change it into a full colour image that looks like it was just taken, and even animate it.

 

 

https://www.facebook.com/share/r/19Zj3pXhXH/

[nomadpete]

Nowadays reality can be created while you wait.

 

And history has become an evolving story, when it was once a record of the past.

[willedoo]

There's been big graphics advances just in the last few months when it comes to image to animated video. Not so long ago people's faces were obviously AI, in looks and facial expressions. With some of the better ones now it's almost spooky how realistic they are becoming. All those little things that define a person's personality are now being captured, especially the eyes and the smaller subtle facial movements.

[willedoo]

In 2005 my brother and I took our dad on a road trip to see the war memorial in Canberra. On the way down we stayed a night in Cowra to have a look at the old POW camp as my dad was in a training battalion there at the time when it happened. In town there was a small museum and they had a hologram show depicting our POW's overseas. The point I'm getting at is that I wonder whether holograms will be replaced by AI created video or whether they will co-exist with it.

[Jerry_Atrick]

6 hours ago, old man emu said:

Simply complying with  Parkinson's Laws: 

In my case, not quite. I am the last to request hiring more people. I am not a natural manager/coach of people and I prefer to do more with less. I was never captain of a team nor raise to any heights at school, etc. I am the literal version of falling upwards. 

 

We genuinely have a lot of work to do. And I am still interviewing people this morning for open roles we need to fill to get work done. 

 

Believe my, I would rather be using AI.. problem is our organisation is extremely risk averse and we are prohibited from using it in an agentic or generative sense for anything other than presentations and the like. But that will not last forever. 

 

Also, much of our workforce is sub contract - one this massive program is over, if it goes to plan, many of the contractors will not be renewed and the sustainable head count will reduce to something like 20 - 30 for ongoing support and maintenance - assuming no AI. 

 

7 hours ago, nomadpete said:

In summary, you haven't listed any personal benefits to the masses.

Like all tech, there are advantages and disadvantages. In theory, once profiteering has been done and firms try to outcompete each other, assuming there is enough competition in the AI space, prices for many items will reduce - and if they are virtual services, massively. Of course, no competition in the AI space will simply drive up AI prices as an input and those cost savings will not flow down. 

 

There are already benefits. I used AI to help with the motorbike - I had to end up sending it to the shop for lack of tools and workshop space (and time), but it did save me quite a bit on labour and also helped me learn more about the bike. I am a terrible artist, but with a simple prompt was able to get AI to design for me exactly what I imagined in my mind for the Halse Angels motorcycle "club":

I could not even come close to producing something like this, even if I had all of my remaining time to do it. 

 

Technological advances in all fields will improve massively - we are seeing early cancer diagnosis that human doctors struggle to spot, remote assisted surgery in remote regions, etc. leading to much better outcomes than previously defined.. All achieved only with the help of AI (or more accurately, much better achieved with the help of AI).  This is a massive benefit to the masses, well except in the USA where universal health care isn't a right. Once autonomous driving kicks in properly, roads will be a lot safer and you will still be able to have your personal bit of real estate on the road when driving. 

 

But of course, there are downsides. AI, as has been seen in the Ukraine war, can result in cheap and super effective lethal force. My view is that the jobs it will create will not be absorbed by the jobs it will displace. And of course, there will have to be a living wage or similar coming from the coffers of somewhere or else there will be civil war. We have seen AI being able to go rogue, but have to remind ourselves these are in !carefully controlled" situations.. i.e. they are not generally the models that are released to the public (even if while the experiment - or more accurately testing is taking place, they may break out of their cages and ho rogue). 

 

Like everything, it is up to humanity to reap the rewards of it. Sadly, the bulk of the "good" humanity is passive and compliant... so when the rogue elements of humanity get in as leadership (look at what we have today and worse, how people will still elect them), AI becomes a dangerous weapon. It is a bit like "Guns don't kill people; people kill people"

 

Even my daughter is now thinking about moving to Aus... somewhere like Finke to get away from it all.

 

(Well, not Finke... but you get the idea).

[nomadpete]

I can see you have tried to answer my question. Thanks for that. But you seem to be struggling to name any definite benefit to the people.

 

What is the prospectus?

 

Will my Council rates/car rego/groceries/rent, etc, go down?

 

Will the hospitals definitely deliver faster/more accurate/cheaper health care?

 

Seems the promises are all couched in terms like 'might' or 'should' do lots of the above.

 

In the past, multibillion dollar projects were only approved when there were specific itemised measurable deliverable values that warrant the investment.

 

So far I only see wishy-washy vague half promises. And little sign of addressing any of the risks.

 

Is it just 'build it and the world will beat a path to your door, bearing gifts of gold' ?

 

 

[Jerry_Atrick]

21 minutes ago, nomadpete said:

Will my Council rates/car rego/groceries/rent, etc, go down?

Councile rates - unlikely - that is a monopoly. That has nothing to do with technology - that is economics and politics. 

Car Rego - same.. Especially if you live in Vic as potholes need fixing.

Groceries - hard to say.. so many factors. AI can greatly assist with all sorts of things to minimise the cost, but if you have a diminishing habitat, increasing input fueld costs, etc. then any gain of AI will be offset by those. 

Rent - almost entirely an economic question of supply and demand.

 

Maybe the question is not if AI will reduce the cost - but what is the difference between the cost with or without AI? I would expect software costs to materially depreciate over time for example as a) more suppliers will come to market through necessity and b) it will be cheaper. Demand should remain relatively constant. 

 

21 minutes ago, nomadpete said:

Will the hospitals definitely deliver faster/more accurate/cheaper health care?

If we use AI in demand predication, medical diagnosis, treatment specification and monitoring and even rostering optimisation, then it should. Again, other economics come into play. We still need all of the tech (scanners, etc) which is very expensive. Will AI help bring down those costs? We still have remote v urban considerations etc. 

 

21 minutes ago, nomadpete said:

Seems the promises are all couched in terms like 'might' or 'should' do lots of the above.

That is an inecitability because it depends on human deployment. The cost with our without AI should be a better gauge. 

 

21 minutes ago, nomadpete said:

In the past, multibillion dollar projects were only approved when there were specific itemised measurable deliverable values that warrant the investment.

 

So far I only see wishy-washy vague half promises. And little sign of addressing any of the risks.

We have many such projects where the cost benefit analysis is called into question during or at the end of delivery. 

 

And the comparison is not quite analogous. If I build a road, it is a product to allow vehicles to travel with safely. It has a specific use, and by itself it does not guarantee safety. It is up to those governing us to put in place the necessary laws and enforcement to ensure practical safety, to stump up the money to maintain the roads, etc. It could be considered a tool for travel, but it is more akin to an end product in itself. 

 

AI is a tool.. it is not the end product in itself. It can be an incredibly general purpose tool such as Chat or Claude or Mistral; or it can be very specific (e.g. autonomous driving AI). Autonomous driving and specific AIs are more like end products in the same way as roads are. And, assume they work, they will bring a massive benefit to the masses.. in easier, safer driving, more efficient driving and hopefully lower ecological costs through that. 

 

The generic AI models are purely tools, list like, I dunno, a jigsaw. In skilled, competent, and conscientious tradespeople, they can produce great things. Sadly, in the hands of cowboys, they can make life quite miserable. 

Edited 2 hours ago by Jerry_Atrick

[rgmwa]

1 hour ago, Jerry_Atrick said:

My view is that the jobs it will create will not be absorbed by the jobs it will displace.

I think that's going to be one of biggest problems it's going to create, although if it manages to wipe out much of humanity along the way, maybe it won't have that much impact.