Ai governmental means-tested energy payments for the low income demographic?

[chrisjones]

The UK government are thoroughly convinced that there is simply no way to means test energy payments to the public in any way for the low-income demographic for example. And that means testing such benefits is too expensive and complex. However, I think that actually, it's very much possible.

As an example, there is a wealth of data held by corporations, banks, and governments on members of the public. My proposal in its simplest form would require the consent from individuals to allow a government or third-party ai system to access this data, such as bank accounts, benefits, employment tax data, etc in order to means test individuals or indeed households and to calculate which individuals are in the low income demographic for the purpose of means-tested energy and other government payments to the public. The ai system would then issue these payments.

Edited September 9, 2022 by chrisjones

[swansont]

That's interesting. I was sure the UK had an income tax, and government assistance programs, which would allow them to see which people have low income.

I would never want my (US) government to be able to tap in to corporate/bank data.

 

And what is your fixation with ai? I don't think it means what you think it means. In this application it's not some magic bullet. It's just computer programming - comparing an income to some predetermined number.

 

[studiot]

23 minutes ago, swansont said:

And what is your fixation with ai? I don't think it means what you think it means. In this application it's not some magic bullet. It's just computer programming - comparing an income to some predetermined number.

+1

But there is more to the issue than this.

The design and implementation of even the simplest computer/software system needs phase of testing on examples with known outcomes.
The greater the complexity of the process being computed the greater the risk of unforseen interactions resulting in unwanted outcomes.
Thus the greater the complexity the greater the need for more wide ranging yet intensive pre implementation testing.

Far too many large 'database' projects have foundered on this rock and AI is no exception.

5 hours ago, chrisjones said:

However, I think that actually, it's very much possible.

So how much testing would be required for you proposal and how long would that last.

How strong would any guarantee of zero to low unwanted outcomes be ?

 

It has been most illuminating (for me) to watch the implementation of computerised robot medicine selection from stock, packaging and labelling, ready for dispensation   -   and to see all the things that can go wrong with even such a limited objective.

[chrisjones]

10 hours ago, studiot said:

+1

But there is more to the issue than this.

The design and implementation of even the simplest computer/software system needs phase of testing on examples with known outcomes.
The greater the complexity of the process being computed the greater the risk of unforseen interactions resulting in unwanted outcomes.
Thus the greater the complexity the greater the need for more wide ranging yet intensive pre implementation testing.

Far too many large 'database' projects have foundered on this rock and AI is no exception.

So how much testing would be required for you proposal and how long would that last.

How strong would any guarantee of zero to low unwanted outcomes be ?

 

It has been most illuminating (for me) to watch the implementation of computerised robot medicine selection from stock, packaging and labelling, ready for dispensation   -   and to see all the things that can go wrong with even such a limited objective.

Hi Studiot a few quick answers.

You said: The design and implementation of even the simplest computer/software system need a phase of testing on examples with known outcomes.
The greater the complexity of the process being computed the greater the risk of unforeseen interactions resulting in unwanted outcomes.
Thus the greater the complexity the greater the need for more wide-ranging yet intensive pre-implementation testing.

My Answer: I agree with these points and also include the observation that UK government IT projects tend to be abject failures both vastly overpriced, over time, clunky, and full of bugs. However, data is the new currency and if the dysfunctional tender system for bidding for government projects could be circumvented I'm sure that one of the new generations of tech companies would welcome this project because data is now so valuable.

You Said: How strong would any guarantee of zero to low unwanted outcomes be?

My answer: I'm reminded of the quote "Don't let the perfect be the enemy of the good" All businesses incur some losses, and in the context of this idea I'm sure that criminal minds would search for loopholes and some people may receive a payment they are not entitled to, however, the current strategy is to pay everyone the same payments rich or poor at a vast cost. But to answer your specific question, then yes testing and failsafes would be required to guard against a vast error of overpayments for example. But the system would also overlook those without a bank account for example, but if every design were to be perfectly inclusive then no design would make it to mark one.

 

10 hours ago, swansont said:

That's interesting. I was sure the UK had an income tax, and government assistance programs, which would allow them to see which people have low income.

I would never want my (US) government to be able to tap in to corporate/bank data.

 

And what is your fixation with ai? I don't think it means what you think it means. In this application it's not some magic bullet. It's just computer programming - comparing an income to some predetermined number.

 

Hi swansot this made me giggle "And what is your fixation with ai? I don't think it means what you think it means. In this application, it's not some magic bullet. It's just computer programming - comparing an income to some predetermined number"

My reply: It does seem I'm obsessed with ai perhaps I think I might be able to have full-blown complex conversations with ai in the near future rather than boring my disinterested family to death. lol

Edited September 9, 2022 by chrisjones

[chrisjones]

On 9/9/2022 at 12:27 PM, studiot said:

+1

But there is more to the issue than this.

The design and implementation of even the simplest computer/software system needs phase of testing on examples with known outcomes.
The greater the complexity of the process being computed the greater the risk of unforseen interactions resulting in unwanted outcomes.
Thus the greater the complexity the greater the need for more wide ranging yet intensive pre implementation testing.

Far too many large 'database' projects have foundered on this rock and AI is no exception.

So how much testing would be required for you proposal and how long would that last.

How strong would any guarantee of zero to low unwanted outcomes be ?

 

It has been most illuminating (for me) to watch the implementation of computerised robot medicine selection from stock, packaging and labelling, ready for dispensation   -   and to see all the things that can go wrong with even such a limited objective.

Hi, studiot to take this discussion further here are a few more thoughts I had on the idea last night. Verification of an individual's identity is often a stumbling block when the UK government intends to issue payments to the public often requiring a photographic id. However many of the public find this difficult for a variety of reasons including but not exclusively the lack of digital resources and technical ability. However much like opening a PayPal account this system would send a small amount of money £0.03 for example to the applicant's bank account that would then require the applicant to enter this figure on the online application form or analog paper application that would consist of tick boxes both online and in paper form. I would suggest that the majority of applicants would have a smartphone to facilitate this process. But would the paper application cause a backlog and require expensive resources to process? The paper forms would be stacked and scanned automatically much like a cash machine counting cash.

[swansont]

20 minutes ago, chrisjones said:

Verification of an individual's identity is often a stumbling block when the UK government intends to issue payments to the public often requiring a photographic id. However many of the public find this difficult for a variety of reasons including but not exclusively the lack of digital resources and technical ability.

How does the government do this now?

 

20 minutes ago, chrisjones said:

However much like opening a PayPal account this system would send a small amount of money £0.03 for example to the applicant's bank account that would then require the applicant to enter this figure on the online application form or analog paper application that would consist of tick boxes both online and in paper form. I would suggest that the majority of applicants would have a smartphone to facilitate this process. But would the paper application cause a backlog and require expensive resources to process? The paper forms would be stacked and scanned automatically much like a cash machine counting cash.

You’re presenting this as a new problem, as if the government doesn’t already do this. Is that actually the case?

[chrisjones]

11 hours ago, swansont said:

How does the government do this now?

 

You’re presenting this as a new problem as if the government doesn’t already do this. Is that actually the case?

Hi swansont I'm going to take a break from posting original or novel ideas but may just comment on other users' threads occasionally as it literally makes my brain hurt lol. As I pointed out in a few of my other threads I have a vasovagal syncope, and for an analogy, my brain is like an old computer that overheats when it is stressed too much.....🤣
🤣🤣🤣👍

[iNow]

No better than a broken bot, then. Sad state that

[exchemist]

2 hours ago, chrisjones said:

Hi swansont I'm going to take a break from posting original or novel ideas but may just comment on other users' threads occasionally as it literally makes my brain hurt lol. As I pointed out in a few of my other threads I have a vasovagal syncope, and for an analogy, my brain is like an old computer that overheats when it is stressed too much.....🤣
🤣🤣🤣👍

But that is fainting, not headaches.

[Ghideon]

15 hours ago, chrisjones said:

to take this discussion further

The discussion includes too many aspects, I would need several pages to adress this. Maybe if you narrow down the topic to some specific issue? There may be interesting connections to ongoing development and research for instance in software architecture. And if you prefer to discuss the "big picture" then you could remove the details that are unnecessary. Example: If your question is how to support all citizens, regardless of their technical abilities and level of access to digital tools, then details about scanners and paper piles may be left out.