A.i in public services

Use this board for general non-cycling-related chat, or to introduce yourself to the forum.
Post Reply
cycle tramp
Posts: 4597
Joined: 5 Aug 2009, 7:22pm

A.i in public services

Post by cycle tramp »

Whilst I disagree with both Elon Musk and Sir Starmer on different things, there's two things that I agree with - in Mr Musk's case, it's that taxes are more likely to be supporting a bureaucracy rather than a democracy and that Sir Starmer is right, one possible way to reduce it is to use A.i.
Having worked the council answering telephones, A.i is already here. Towards to later part of my employment with the council, our management had invested in a computer modelling system which predicted the number of calls per day, when they would occur and the type of calls.. management could then set our shifts to make the best use if the time for which we were being paid... at the start it was a bit hit and miss, but due to various feedback loops within the system it got cleverer and cleverer.. the upshot being that the council never paid for any customer call handers that it didn't need.
At the moment lots of council officers are being paid to make judgement calls on the evidence provided by customers.. these include, which vulnerable people need to be housed, whether certain benefits should be awarded, whether council tax or nndr be awarded, whether parking fines should be upheld or quashed... some of these decisions will be complicated but a great many more won't be. But they tie up alot of officer time, which means cost to the tax payer..
Cont'd
Last edited by cycle tramp on 13 Jan 2025, 8:48pm, edited 3 times in total.
Dedicated to anyone who has reached that stage https://m.youtube.com/watch?v=Vqbk9cDX0l0 (please note may include humorous swearing)
cycle tramp
Posts: 4597
Joined: 5 Aug 2009, 7:22pm

Re: A.i in public services

Post by cycle tramp »

An example of this was the benefits team which, when I left work consisted if a manager and about 12 members of staff, being padded between 24 - 35k each year, each year
Local services desperately needs to be cheaper, and make more efficient use of your council tax.
That's not to say we won't need benefit officers to make decisions, but their time would be slot better focused on the complicated claims.
A.i not only represents a method to achieve this, it should also ensure a consistent customer experience across the whole of council services and should also ensure that no vital evidence is lost or misunderstood. Its even possible that A.i. could go out and search for evidence, which could make things like benefit claims less open to fraud.. it should also allow for decisions to be made faster. After all, you can run the A.i overnight, over the weekend and even over Bank Holidays.

Generally I have been known to frown at new things (tubeless tyres, disc brakes, carbon fibre) but not this time -A.i could pave a way to a better, more efficient, more cost effective future, more equitable future, it could even be used to start to predict and tailor council services to the needs of society and the individual.. compared to the 're-active' state of our current services, this may be a good thing.

An example of this would be A.i modelling which looks at the number of children born in any one area, those children who have moved in and calculates school placement numbers, the current capacity and then flags any warns to those involved...
The same prediction service could look at the provision of care homes...
The right A.i system may be able to predict what the rise of council tax may cause to the local economy.

What A.i gives us, is perhaps for the first time, the possibility to predict where the thunderstorm will form, how much rain will fall and how long it will last for when the butterfly flaps its wings
Dedicated to anyone who has reached that stage https://m.youtube.com/watch?v=Vqbk9cDX0l0 (please note may include humorous swearing)
User avatar
simonineaston
Posts: 8815
Joined: 9 May 2007, 1:06pm
Location: ...at a cricket ground

Re: A.i in public services

Post by simonineaston »

I read that AI can whizz through huge quantities of images, correctly identifying elements of interest at a rate far higher than any radiologist. Never tires, never has a bad day - sounds good to me!
S
(on the look out for Armageddon, on board a Brompton nano & ever-changing Moultons)
PDQ Mobile
Posts: 4911
Joined: 2 Aug 2015, 4:40pm

Re: A.i in public services

Post by PDQ Mobile »

I can't see that AI with it's associated cost is more effective at predicting when peak times on a help line or other council service will occur.
A simple manual look at the records or even, heaven forbid, ask the people concerned will do the same job.
It's not rocket science.

Additionally you say the council "never paid for any call handlers it didn't need" but that arguably led to longer waiting times for the public trying to phone in.
Can't see it could be otherwise.
.........

On the related theme of Ai fixed potholes.
There is locally a small housing estate that uses a small steep road as access.
This section of the steep road has ground water flowing down it- since forever.
All the infrastructure is in place to contain and catch this water but leaves block the main grid adjacent to the estate access point and as a consequence it becomes seriously icy across the whole road downhill in freezing conditions.
Every day in such conditions the council send a lorry to salt the section and every day the water washes the salt away and so on and so forth.
It is the economics of the madhouse.
(Road sweepers are rare as hen's teeth here)

Neither the council ( or any of the residents) have the nouse or inclination to spend 5 minutes using a shovel and broom to remedy the issue.

No amount of AI will solve this kind of problem.
It is a simple problem that needs a simple solution.
Nearholmer
Posts: 5627
Joined: 26 Mar 2022, 7:13am

Re: A.i in public services

Post by Nearholmer »

I’m not by any means sure that many of the above examples need or amount to AI, they sound like pretty basic algorithms that could be encoded behind, for instance, a simple spreadsheet, and probably already are.

Which begs the question “what is AI?”, or maybe “where does the boundary lie between conventional, deterministic programming and AI?”

This is one definition of AI, but doesn’t seem very helpful:

“The theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages”

“Decision making” for instance is a very wide field, and a simple photo-cell circuit to turn a street light on and off might be called “artificial intelligence” by this definition, since it decides whether the criteria for “dark enough” and “light enough” have been met.
User avatar
[XAP]Bob
Posts: 20231
Joined: 26 Sep 2008, 4:12pm

Re: A.i in public services

Post by [XAP]Bob »

cycle tramp wrote: 13 Jan 2025, 8:24pm Whilst I disagree with both Elon Musk and Sir Starmer on different things, there's two things that I agree with - in Mr Musk's case, it's that taxes are more likely to be supporting a bureaucracy rather than a democracy and that Sir Starmer is right, one possible way to reduce it is to use A.i.
What is democracy if not bureaucracy?
We need those civil servants to be doing their jobs.

AI is crap at basically everything at the moment - it has very specific use cases where ML is fantastic, radiography images being one.
It's not going to solve major problems of humanity any time soon...
A shortcut has to be a challenge, otherwise it would just be the way. No situation is so dire that panic cannot make it worse.
There are two kinds of people in this world: those can extrapolate from incomplete data.
PDQ Mobile
Posts: 4911
Joined: 2 Aug 2015, 4:40pm

Re: A.i in public services

Post by PDQ Mobile »

Nearholmer wrote: 14 Jan 2025, 11:00am I’m not by any means sure that many of the above examples need or amount to AI, they sound like pretty basic algorithms that could be encoded behind, for instance, a simple spreadsheet, and probably already are.

Which begs the question “what is AI?”, or maybe “where does the boundary lie between conventional, deterministic programming and AI?”

This is one definition of AI, but doesn’t seem very helpful:

“The theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages”

“Decision making” for instance is a very wide field, and a simple photo-cell circuit to turn a street light on and off might be called “artificial intelligence” by this definition, since it decides whether the criteria for “dark enough” and “light enough” have been met.
Yes let's see it well defined before we go down the road.
PDQ Mobile
Posts: 4911
Joined: 2 Aug 2015, 4:40pm

Re: A.i in public services

Post by PDQ Mobile »

[XAP]Bob wrote: 14 Jan 2025, 11:54am
cycle tramp wrote: 13 Jan 2025, 8:24pm Whilst I disagree with both Elon Musk and Sir Starmer on different things, there's two things that I agree with - in Mr Musk's case, it's that taxes are more likely to be supporting a bureaucracy rather than a democracy and that Sir Starmer is right, one possible way to reduce it is to use A.i.
What is democracy if not bureaucracy?
We need those civil servants to be doing their jobs.

AI is crap at basically everything at the moment - it has very specific use cases where ML is fantastic, radiography images being one.
It's not going to solve major problems of humanity any time soon...
ML?
Manual labour?
the snail
Posts: 416
Joined: 5 Aug 2011, 3:11pm

Re: A.i in public services

Post by the snail »

Nearholmer wrote: 14 Jan 2025, 11:00am ...

“The theory and development of computer systems able to perform tasks normally requiring human intelligence, such as visual perception, speech recognition, decision-making, and translation between languages”

“Decision making” for instance is a very wide field, and a simple photo-cell circuit to turn a street light on and off might be called “artificial intelligence” by this definition, since it decides whether the criteria for “dark enough” and “light enough” have been met.
Actually that is a pretty good definition. The street light example doesn't meet the criterion, since it doesn't require intelligence, but could be controlled by a procedural program implementing a formula. Something like speech recognition is much more complex, and afaik, impossible with traditional approaches. AI encompasses a wide range of techniques, but generally they are mostly useful where the data is messy and the decision making process is fairly opaque.
User avatar
[XAP]Bob
Posts: 20231
Joined: 26 Sep 2008, 4:12pm

Re: A.i in public services

Post by [XAP]Bob »

PDQ Mobile wrote: 14 Jan 2025, 2:39pm
[XAP]Bob wrote: 14 Jan 2025, 11:54am
cycle tramp wrote: 13 Jan 2025, 8:24pm Whilst I disagree with both Elon Musk and Sir Starmer on different things, there's two things that I agree with - in Mr Musk's case, it's that taxes are more likely to be supporting a bureaucracy rather than a democracy and that Sir Starmer is right, one possible way to reduce it is to use A.i.
What is democracy if not bureaucracy?
We need those civil servants to be doing their jobs.

AI is crap at basically everything at the moment - it has very specific use cases where ML is fantastic, radiography images being one.
It's not going to solve major problems of humanity any time soon...
ML?
Manual labour?
Machine learning - a far more accurate descriptor of what we currently have than AI. It's all just advanced pattern matching, no intelligence involved.
A shortcut has to be a challenge, otherwise it would just be the way. No situation is so dire that panic cannot make it worse.
There are two kinds of people in this world: those can extrapolate from incomplete data.
Nearholmer
Posts: 5627
Joined: 26 Mar 2022, 7:13am

Re: A.i in public services

Post by Nearholmer »

The street light example doesn't meet the criterion, since it doesn't require intelligence, but could be controlled by a procedural program implementing a formula
No difference. The light controller is following a formula, a tiny piece of logic. If the lights were controlled by a procedural formula, the controller would possibly need some more sensors/inputs (time maybe, number of pedestrians on the street etc), and maybe a few more logic steps, but it wouldn’t be doing anything materially different. In fact, the earliest street lighting procedural programmable controllers were nothing but an analogue clock, with a disc and pins that could be set to actuate contacts, energizing relays/contactors, so still only one input.

I still think the definition is poor, because it starts with the very wobbly phrase “normally requiring human intelligence”. Every single task ever that was previously performed by a human and was then given to a machine might be said to fall in that bracket, in that it normally required human intelligence until it was given to the machine.

Another definition, which I think might be more useful:

“Artificial intelligence is a field of science concerned with building computers and machines that can reason, learn, and act in such a way that would normally require human intelligence or that involves data whose scale exceeds what humans can analyze. ”

“Learn” and “scale exceeds what humans can analyse”, seem to me to be important in differentiating AI as we commonly understand it from what went before.
Nearholmer
Posts: 5627
Joined: 26 Mar 2022, 7:13am

Re: A.i in public services

Post by Nearholmer »

Here’s something to watch:

https://m.youtube.com/watch?v=_FzqIZe2dyQ

This chap is the father of one of my son’s pals, and to say that he knows a lot about AI would be a gigantic understatement.
UpWrong
Posts: 2782
Joined: 31 May 2009, 12:16pm
Location: Portsmouth, Hampshire

Re: A.i in public services

Post by UpWrong »

I recently applied for research funding to predict nursing staff requirements in hospital wards shift-by-shift using routinely recorded (and anonymised) patient data. My proposal suggested considering prediction models using both classical regression methods and deterministic machine learning methods. The reviewers wanted me to have patient and public involvment in the study to understand the acceptability of using AI in healthcare. I had deliberately avoided using the term "AI" in the proposal. The Health Foundation has run a recent survey on this with some useful results which suggests acceptability does depend on what the application is and having safeguards, https://www.health.org.uk/reports-and-a ... taff-think
Stevek76
Posts: 2213
Joined: 28 Jul 2015, 11:23am

Re: A.i in public services

Post by Stevek76 »

Nearholmer wrote: 15 Jan 2025, 7:43am Here’s something to watch:

https://m.youtube.com/watch?v=_FzqIZe2dyQ

This chap is the father of one of my son’s pals, and to say that he knows a lot about AI would be a gigantic understatement.
For me that overstates matters in the other direction to the consistent hype from the AI tech bros.

There's no scientific basis for any mechanism that makes the human brain (or indeed those of the more intelligent animals) some uniquely unreproducible and that we aren't simply following the internal logic of our brains and biological systems. Though that really wanders into to the thorny realms of free will and what consciousness actually is.
The contents of this post, unless otherwise stated, are opinions of the author and may actually be complete codswallop
User avatar
[XAP]Bob
Posts: 20231
Joined: 26 Sep 2008, 4:12pm

Re: A.i in public services

Post by [XAP]Bob »

UpWrong wrote: 15 Jan 2025, 10:40am I recently applied for research funding to predict nursing staff requirements in hospital wards shift-by-shift using routinely recorded (and anonymised) patient data. My proposal suggested considering prediction models using both classical regression methods and deterministic machine learning methods. The reviewers wanted me to have patient and public involvment in the study to understand the acceptability of using AI in healthcare. I had deliberately avoided using the term "AI" in the proposal. The Health Foundation has run a recent survey on this with some useful results which suggests acceptability does depend on what the application is and having safeguards, https://www.health.org.uk/reports-and-a ... taff-think
"I had deliberately avoided using the term "AI" in the proposal."
Well done, describe what you're using, not some marketing drivel.
A shortcut has to be a challenge, otherwise it would just be the way. No situation is so dire that panic cannot make it worse.
There are two kinds of people in this world: those can extrapolate from incomplete data.
Post Reply