the_upsetter
Posts: 4
Joined: Thu Jan 16, 2020 4:00 pm

algorithms and ethics at the Barbershop

Thu Jan 16, 2020 4:20 pm

after reading through some of this intresting post on the blog ...

https://www.raspberrypi.org/blog/can-al ... -unethical -
it is important that we have these conversations, especially because people often see computers as ‘magic boxes’
i first recalled reading about https://plus.maths.org/content/mathemat ... rs-paradox

... hear some of my (aparently to ridiculous for the blog-comments) thaughts about these important conversations
wrt > content algorithms

“content algorithms observe the choices you make and the content you consume.”

afaik an algorithm has no point to observe from

a human might write an algorithm that infers things
about what *they* beleave are *your* choices/content.

> The most effective algorithms give you more of the same content, with slight variation.

this appears to make the assumption

we all catogorise content in the same way
the algorithm writer *catogorises content* as “the same”

not just that !
but it also assumes

*we all catogorise content in the same way
as the algorithm writer*
catogorised
the content
at the exact time
they wrote the algorithm

Heater
Posts: 14240
Joined: Tue Jul 17, 2012 3:02 pm

Re: algorithms and ethics at the Barbershop

Thu Jan 16, 2020 6:57 pm

the_upsetter,

I totally fail to see any connection between Russell's paradox and the use of AI/deep learning or other algorithms. Perhaps you can explain.

I'm glad Helen Joyce said her comment about the sex of the barber was flippant because it is pretty stupid. I'm sure there are female barbers and I'm sure there are girls that shave. Besides, logically it makes no sense, if the original paradox was about the set of men and which shave which, by introducing a third set, of girls, one has totally changed the problem.

I don't want to talk about ethics and morals much. Theologians, philosophers and others have been arguing about that for centuries, with no definitive conclusion yet.

However I would like to comment on the questions posted in that blog post:

- What algorithms do you interact with every day?

I have no idea. How can I know? Of course my computer is teeming with algorithms, for all kind of things in the operating system and applications. Everything from Euclid's two thousand year old GCD to the Cooley–Tukey FFT algorithm and Tony Hoare's Quick Sort of the 1960's. And many more besides.

Of course as soon as I hit the net I'm dealing with Google's Page Rank algorithm and all kind of AI/deep learning etc that I have no idea about.

- How large are the decisions you allow algorithms to make?

This is a dumb question. I have no control over the size of decisions those algorithms make. Out there on the net, used by other people to make decisions about my life. For example:

My bank may decide to loan me money or not. Depending on what their algorithms tell them about my financial history and whatever else they can find.

Insurance companies will adjust my insurance rates depending on what their algorithms tell them about my medical history, my driving record and whatever else they can find.

Companies may decide to employ me or not. Depending on what their algorithms tell them about me.

And so on, and so on, more and more everyday.

- Are there algorithms you absolutely do not trust?

See above. I can trust the likes of Euclid's GCD and Tony Hoare's Quick Sort. I cannot trust anything I do not know even exists or where it is or who is using it for what purpose or how it works.

- What do you think would happen if we let algorithms decide everything?

Interesting question. We seem to be performing the experiment to find out as we speak. On a global basis.

TL;DR

Algorithms do not have ethics or morals. The people who use them as tools to get what they want done might have (Although the latter is not always clear to me)
Memory in C++ is a leaky abstraction .

User avatar
Burngate
Posts: 6147
Joined: Thu Sep 29, 2011 4:34 pm
Location: Berkshire UK Tralfamadore
Contact: Website

Re: algorithms and ethics at the Barbershop

Thu Jan 16, 2020 7:31 pm

Heater wrote:
Thu Jan 16, 2020 6:57 pm
Companies may decide to employ me or not. Depending on what their algorithms tell them about me.
That's been going on since time immemorial. They were just using their own inbuilt biases instead of the ones built into their algorithms.

I was offered an interview for a job near London, that happened to coincide with my holiday in Scotland.
I could have declined the interview, or cancelled my holiday, but instead chose to drive to Glasgow, fly to Heathrow, attend the interview, and fly back. Only one day of my holiday lost.
I happened to mention that, and I'm convinced they gave me the job because they were so impressed by my enthusiasm. They probably regretted that for the next twenty years.

As for the Barber's problem, the implicit built-in assumption is that all men must shave or be shaved

Heater
Posts: 14240
Joined: Tue Jul 17, 2012 3:02 pm

Re: algorithms and ethics at the Barbershop

Thu Jan 16, 2020 8:23 pm

Burngate,
That's been going on since time immemorial. They were just using their own inbuilt biases instead of the ones built into their algorithms.
I'm very sure that is true.

How much do we want to mechanize and amplify biases with our computer power? How much do we want to give that mechanized and amplified power to very few actual humans?

Call me old fashioned but back in the day one would make an appointment with ones local bank manager, for example, he would invite you into his office and discuss your proposals and/or problems over a coffee. And offer advice. If he turned you down for a loan you likely knew why.

Today, there is no human to speak to unless you are a big customer I guess. Decisions are made anonymously through web interaction with some "algorithm". There is no recourse. There is no one to talk to.

I like your job hunting anecdote, I could tell similar stories. That was apparently more than 20 years ago. Today they might have you on a video conference call where an AI thing evaluates your facial expressions and responses. Assuming an algorithm did not reject your CV immediately for reasons unknown.

Yes, the Barbershop problem is a crude human language analogy to the set theoretic issue that Russel raised at the heart of mathematics.
Memory in C++ is a leaky abstraction .

the_upsetter
Posts: 4
Joined: Thu Jan 16, 2020 4:00 pm

Re: algorithms and ethics at the Barbershop

Fri Jan 17, 2020 11:25 am

Burngate wrote: As for the Barber's problem, the implicit built-in assumption is that all men must shave or be shaved
Heater wrote:
Thu Jan 16, 2020 8:23 pm
Yes, the Barbershop problem is a crude human language analogy to the set theoretic issue that Russel raised at the heart of mathematics.
perhaps it wa/is a bit of an obtuce reffrence to Occam's razor ;)

and related theoretical topiary :)
one might encounter while strolling along this particular mathematical garden path.

hippy
Posts: 6500
Joined: Fri Sep 09, 2011 10:34 pm
Location: UK

Re: algorithms and ethics at the Barbershop

Fri Jan 17, 2020 1:41 pm

The algorithm which intrigues me, and is Pi related, is whatever CeX are using to set their pricing for second-hand Pi's, and probably everything else they sell.

The price of a Zero W has shot up from £10, to £12, then £15, and now £18, over the course of just a few months. The Pi 4B 1GB has though dropped from £45 to £30, 4GB down from £65 to £50.

Prices for other Pi versions don't rationally reflect a board being better than another so I can only guess it's based on rarity and demand, perhaps how quickly they sell when they come into stock, or maybe from what people are searching for, product pages they are viewing.

The ratio of prices they pay against what they sell at also seems bizarre. They'll pay £4 cash for a Zero, A or B which they sell at £10, yet only £2 for a 2B they sell at £15. I would guess it also comes down to demand and rarity, how long it's likely to sit on the shelf for.

It's an algorithm which affects me because I'm in the market for cheap second-hand Pi's, but I can't say if it's bad or good, ethical, moral, or not. It simply is what it is. Sometimes it benefits me, sometimes it doesn't. I'm happy to buy B+'s at almost half the price of a Zero W, 3B's at not much more, but I'm not buying any more Zero W's until the price drops.

I suppose there may be an issue if they are setting prices based on what I and others want, what we are checking the prices for, but it's hard to fault a company for seeking to maximise profits. If I consider it value for money, cheaper than what I would pay elsewhere, it's a win when I buy it. Otherwise I don't.

And it's the same price everywhere, not based upon who one is or where they are. That would be more problematic which is where the concern over 'algorithms' mostly lies.

the_upsetter
Posts: 4
Joined: Thu Jan 16, 2020 4:00 pm

Re: algorithms and ethics at the Barbershop

Sat Jan 18, 2020 10:01 am

Heater wrote:- What algorithms do you interact with every day?
would you believe!
to signup to this verry forum
i had to train some ML
or "AI"
if *you* like to put it in the set of intelligent things

one example techradar.com/*/captcha-if-you-can-how-youve-been-training-ai-for-years-without-realising-it

this is something i have heard some discussion of wrt to the ethical issues around humans training algorithms

and another : 'When AI needs a human assistant - Human workers are training algorithms behind the scenes'- .theverge.com/*/amazon-mturk-google-captcha-robot-ai-artificial-intelligence-mechanical-turk-humans

which some might argue calling algorithm trained in this way "artificial" and "intelligence" is alittle disingenuous :!:

it even gets a mention in the FT : .ft.com/ The humans behind Mechanical Turk’s artificial intelligence - Highly educated workers drive machine learning for Amazon by completing tasks


i guess the *real* progress is being made by advertising and PR ?

as this topic of "digital-intelligence" has croped up again in last years christmas lectcures
- ai-and-law-expert-to-feature-in-royal-institutions-christmas-lecture

i recalled a previous lecture talking about categorising animals , in this case it was a Cat or Dog challenge

rigb.org/christmas-lectures/watch/2008/hi-tech-trek/digital-intelligence
- i just saw that a Lecture transcript is avalable in the case it might be of use to anyone reading

+ - Digital intelligence activities to try at home - https://www.rigb.org/christmaslectures0 ... m#lecture5

have a good weekend !

Heater
Posts: 14240
Joined: Tue Jul 17, 2012 3:02 pm

Re: algorithms and ethics at the Barbershop

Sat Jan 18, 2020 11:11 am

the_upsetter,
would you believe! to signup to this very forum i had to train some ML or "AI" if *you* like to put it in the set of intelligent things
What, only now have people realized what is going on with the recaptcha things? We have never known for sure but it has been discussed around our office for years now.

Oddly enough I have never had to do any word picking to log into anything. Usually little tiled pictures of streets and the task is to pick all the tile that contain a bridge or road traffic light or some such.
..which some might argue calling algorithm trained in this way "artificial" and "intelligence" is alittle disingenuous
Hmm...

I was always unhappy about the term "artificial intelligence". After all if it is actually intelligent then it's 'intelligence"

Think about the word "computer". Originally "computers" were human beings. "computer" was an occupation. Hoards of people, usually women, were employed to do arithmetic all day, performing large calculations on artillery trajectories or navigation tables etc. When the electronic computer arrived and replaced all those women computers it was known as the 'electronic computer'. Today we don't expect humans to do that kind of tedious computing so now we expect 'computer' to refer to 'electronic computer'

I could speculate that at some point in the future we will not expect or need to employ any actual humans to use their intelligence to do all that tedious thinking. At which point 'artificial intelligence' will become just 'intelligence' and there will be no ambiguity.

Of course you seem to be assuming that whatever goes on in the neural networks of humans brains when they learn and recognize things is somehow fundamentally different to what goes on when we train networks of artificial neurons. Thus actual human intelligence is somehow different from 'artificial intelligence' and hence the latter term is disingenuous. This is not so clear to me.
Memory in C++ is a leaky abstraction .

User avatar
Gavinmc42
Posts: 4201
Joined: Wed Aug 28, 2013 3:31 am

Re: algorithms and ethics at the Barbershop

Sat Jan 18, 2020 12:08 pm

Today they might have you on a video conference call where an AI thing evaluates your facial expressions and responses
With telecommuting these days how long before you question the person you hire is not an AI bot?
One person with a bot farm could be getting paid to do ten jobs?

Could a Pi with Darknet/yolo be able to get past the recaptcha thing?
I thought learning Linux on Pi's was going to be it for awhile.
I did not expect to be doing AI so soon, but that's what I was doing over the xmas break.
Pi4's are good enough now.

I saw a video of a garage startup the other day, a robot burger machine.
Menial burger flipping jobs will disappear faster than we can expect?
Now thinking about my kids first job in another light.

Logical paradoxes have been used to short circuit robot brains in stories.
The hero saves the world etc. Was that "War Games"?

Ethical paradoxes?
Time to reread Asimov Foundation series?
And the Robot series?

I can see "The Spike" getting closer , not sure which will get here faster, the Technical Singularity or Fusion Power?
I suspect the first will need the second.
While bots still have batteries we should be safe.

Decisions made/assisted by AI, will they be better?
Define "Better", define Ethics?
Who makes better decisions Kirk or Spock.
We know who make the more logical ones.
Will AI's make more logical decisions or will they surprise us?

Decade ago when i came across robots, Tilden's Beam bots and Brooks Subsumption coded ones.
The term "Emergent behaviour" was coined?
Will we see Emergent Ethics?

Anyone know what is the best computer language for Ethical coding?
I'm dancing on Rainbows.
Raspberries are not Apples or Oranges

the_upsetter
Posts: 4
Joined: Thu Jan 16, 2020 4:00 pm

Re: algorithms and ethics at the Barbershop

Sat Jan 18, 2020 11:15 pm

Heater wrote: At which point 'artificial intelligence' will become just 'intelligence' and there will be no ambiguity.
> "At which point" .. from who's perspective ? ... intrestingly russell get a few mentions in this lecture :roll:

perhaps the ambiguity will simply dissapear when the "digital-intelligence" is implemented in a `quantum computer` :?: :|

User avatar
Gavinmc42
Posts: 4201
Joined: Wed Aug 28, 2013 3:31 am

Re: algorithms and ethics at the Barbershop

Sun Jan 19, 2020 1:32 am

perhaps the ambiguity will simply disappear when the "digital-intelligence" is implemented in a `quantum computer` :?:
Nope that will just mean they will be ethical and unethical at the same time.
I have an interest in AI and Law.
While laws are black and white, interpretations of the laws and sentencing is very grey.
This grey, fuzzy area reminds me of Fuzzy logic.
Playing around with FPGA tools on Pi's, I started looking at ternary logic -1, 0, 1.

Yes, No , Maybe.
Looking for a definitive answer, Yes, No to questions of behaviour I don't feel allows for those grey areas.
In Law the Jury system relies on the combined life experiences of the Jury.
You could say those life experiences are the equivalent of human behaviour algorithm training.
But it still works under certain rules, innocent till proven guilty, no conviction just based on one persons word....
The beyond reasonable doubt is a rule for a black and white answer.

Ethics, law, politics are areas that could use AI, just because lots of people get affected by those grey areas.
Blackstone's formulation is a rule to let someone out of the grey area, an escape key so to speak.

There are rules in law, some are unwritten or buried in legislation.
What are the rules for ethics?
Democracy is supposed to be based on the rule of law.
Ethics in a democracy is supposedly held to higher standard, does that mean better rules of law?

If you make an AI that upholds the rule of law, do you get an ethical AI?
Could anyone tell the difference?
Most people have an inbuilt radar for Natural Justice, at least in most democracies and probably other countries too.
What happens when people trust the AI's more than humans when making decisions?
Would that still be a democracy?
Would people vote for the best AI?
Have a jury system of different AI algorithms that reach consensus?

Different Cultures would insert different bias into their AI bots?
Open source the AI's and the best AI's rule the world?
AI decisions are being made about you all the time, every time you click on something on the net...
Are they making ethical decisions for what ads they show you?
I'm dancing on Rainbows.
Raspberries are not Apples or Oranges

User avatar
Douglas6
Posts: 4800
Joined: Sat Mar 16, 2013 5:34 am
Location: Chicago, IL

Re: algorithms and ethics at the Barbershop

Sun Jan 19, 2020 1:56 am

'Algorithm' is simply a modern term. They have been employed for thousands of years. They have become faster and more complex with the aid of computers.

'AI' is just a bastardization of 'pattern recognition'. Non-human intelligence is not even on the horizon. Wake me when a computer can tell me what it's like to see red.

ejolson
Posts: 4030
Joined: Tue Mar 18, 2014 11:47 am

Re: algorithms and ethics at the Barbershop

Sun Jan 19, 2020 5:13 am

Douglas6 wrote:
Sun Jan 19, 2020 1:56 am
'Algorithm' is simply a modern term. They have been employed for thousands of years. They have become faster and more complex with the aid of computers.

'AI' is just a bastardization of 'pattern recognition'. Non-human intelligence is not even on the horizon. Wake me when a computer can tell me what it's like to see red.
As dogs, cats and other animals seem to experience emotions quite distinctively, it seems plausible that deep-learning convolutional neural networks will develop emotions far sooner than logic and self awareness. As a result, the chances of a computer seeing red within the next decade seem very probable to me. A similar idea framed in an evolutionary context may be found at

https://www.independent.co.uk/news/scie ... 94815.html

At the same time, the colour-blind bull in the arena will clearly tell you that seeing red is very different than being able to explain what it's like.

User avatar
Gavinmc42
Posts: 4201
Joined: Wed Aug 28, 2013 3:31 am

Re: algorithms and ethics at the Barbershop

Sun Jan 19, 2020 6:20 am

Wake me when a computer can tell me what it's like to see red.
You do realise you were trained to see red?
Probably before you could talk?
Some people can taste colours.
Is red different for people who see to those that taste it if they both agree it is red?

If you have an algorithm that learns, what is then the algorithm?
While playing with CV using Darknet/Yolov3, the training came in a weighted 238MB file.
The algorithm just used that file.
So is it the trained file or the algorithm that identifies objects?

For ethics will it be the algorithm or the weighted training data, assuming it follows a similar model to CV.
I'm dancing on Rainbows.
Raspberries are not Apples or Oranges

Heater
Posts: 14240
Joined: Tue Jul 17, 2012 3:02 pm

Re: algorithms and ethics at the Barbershop

Sun Jan 19, 2020 10:20 am

Douglas6,
'Algorithm' is simply a modern term. They have been employed for thousands of years. They have become faster and more complex with the aid of computers.
'Algorithm' is certainly not a modern term. It's been in use for a couple of thousand years. I fear that 'algorithm' is one of those technical terms from mathematics and CS that is now being bandied about a lot by those who have little idea what one is. Euclid would be turning in his grave. Not to mention Muḥammad ibn Mūsā al-Khwārizmī from whose name the word is derived.
'AI' is just a bastardization of 'pattern recognition'. Non-human intelligence is not even on the horizon.
Intuitively my gut tells me you are right. But I can't put my finger on a sound logical reason why not. It's not clear to me that whatever humans have is any more than 'pattern recognition'. All be it on a much larger scale with more sophistication.
Wake me when a computer can tell me what it's like to see red.
I have a challenge for you. Can you tell me what it is like to see the colour red? Or did you mean 'see red' as as in becoming very angry, if so you can tell me what that is like instead.

Having done so, if you can, if I had an AI that produced the same or similar answer as you do, how should I make a distinction between you, presumably human, and my machine?
Memory in C++ is a leaky abstraction .

Heater
Posts: 14240
Joined: Tue Jul 17, 2012 3:02 pm

Re: algorithms and ethics at the Barbershop

Sun Jan 19, 2020 10:47 am

Gavinmc42,
Nope that will just mean they will be ethical and unethical at the same time.
That is the most profound and yet witty statement.
You do realise you were trained to see red?...Probably before you could talk?
I don't believe that is true.

One of my earliest memories is crawling across a room toward a beautiful, bright, strange thing. When I got closed I was fascinated by it, nothing else looked like that. So I reached out and shoved my hand at it. All of a sudden there was searing pain in my fingers. That thing was glowing red. It was an element of an electric bar heater!

In that moment I learned a lot. About heat and light and pain. I suspect I did not have words for any of them at the time. That color red is still burned into my memory.
So is it the trained file or the algorithm that identifies objects?
Well there is the thing. The algorithm is the same in all cases. But my it's training it can be used to classify things in all kind of different ways. To recognize or not recognize things in very different ways.
For ethics will it be the algorithm or the weighted training data, ..
I would argue neither. Let me make the point by example:

I could train an algorithm to recognize people and faces. I could further train it with thousands of examples of people I label as 'employable' or 'not employable'. I could then hire people or not depending on what my 'algorithm' says of them. That saves me having to check all new applicants myself. It save me having to hire an HR department to do that checking for me. Great!

But wait you say. That is not fair. That is not ethical. Your algorithm on selects as employable young beautiful people of fair skin. It rejects bald and fat people. And so on.

Well of course it might. Because that is how I trained it. I will have put my bias in there, even unintentionally. Perhaps I thought I was classifying 'intelligent' looking people. Which I perhaps unwittingly conflate with "young beautiful people of fair skin".

The ethics is not in the algorithm or the data. It's in those who build and use it.
Memory in C++ is a leaky abstraction .

User avatar
Gavinmc42
Posts: 4201
Joined: Wed Aug 28, 2013 3:31 am

Re: algorithms and ethics at the Barbershop

Sun Jan 19, 2020 11:18 am

The ethics is not in the algorithm or the data. It's in those who build and use it.
Those who build it and train it might not be those who use it.
We have an issue in my country with a Uni who had a visiting prof.
They got money to make AI vision that is now suspected being used in the Prof's original country.

Is what we are talking about building Ethics into the AI so it cannot be misused?
Good luck with that.

As for using AI's for employing people just ask Google, it knows everything.
Jussie Smollett is just about to find out how much Google knows about him.

Bias and training for people selection?
That's why we run tests of knowledge and have a diverse selection board.
The same thing could be done with AI.
No reason just to have one AI/algorithm, use many.
Can you tell me what it is like to see the colour red
That is not the same as identifying a colour, that is emotion.
Do we need emotional robots?
Just ask Marvin.
Did not see Star Wars, but I thought I saw a short clip where C3PO had emotions?

I have been using i2c AMS colour sensors for years, that gives me 16bits of RGB.
I then used 6 colour ones. The new ones are now 11 colours.
It would easy to make a robot arm with colour sensor that points to an object and says "red".
Plenty of same thermal sensor, points and says "hot".
How much training is needed for that?

Heater, is that how you got your handle?
Single lesson training program, ouch.
I'm dancing on Rainbows.
Raspberries are not Apples or Oranges

Heater
Posts: 14240
Joined: Tue Jul 17, 2012 3:02 pm

Re: algorithms and ethics at the Barbershop

Sun Jan 19, 2020 12:02 pm

Gavinmc42,
Heater, is that how you got your handle?
You know, as I wrote that the same question came into my mind. And it was a rather disturbing thought.

I always thought I chose the handle 'heater' because many of my first ever electronics projects used devices like these:

Image

Which as you know have heaters.

Perhaps I did not choose my handle at all. Perhaps my neural net found it for me as a result of that early training.
Memory in C++ is a leaky abstraction .

Heater
Posts: 14240
Joined: Tue Jul 17, 2012 3:02 pm

Re: algorithms and ethics at the Barbershop

Sun Jan 19, 2020 12:26 pm

Gavinmc42,
That is not the same as identifying a colour, that is emotion.
Of course. Emotion, feeling, being aware, consciousness, the experience....

I was not asking about any that.

The challenge was: Can you tell me what it is like to see the colour red?

You might respond to that with all kind of words. I can only get your written words here. You can say anything you like about it. As you did. Other people might say somewhat different things.

Well now, I can imagine an AI that produces very similar results. Perhaps it has been trained on the entire text of the internet and can concoct a similar response. Perhaps it can even link me to images that contain red, just so I know what it is talking about. It might even talk about emotions, as you do.

How would I tell the difference between it's intelligence and yours?

How would I know, from the responses, it feels anything, has any emotions, awareness, consciousness, whatever you want to call it, or not.

More importantly, how would I know, from the responses, that you do? No matter how you protest that you do have emotions and that makes you some how different from my AI. How can I tell the difference?

All in all, I don't think the "Wake me when a computer can tell me what it's like to see red" thing is much of a good criteria.
Memory in C++ is a leaky abstraction .

User avatar
rpdom
Posts: 15882
Joined: Sun May 06, 2012 5:17 am
Location: Chelmsford, Essex, UK

Re: algorithms and ethics at the Barbershop

Sun Jan 19, 2020 12:32 pm

Heater, something like this then?
pivalve.jpg
pivalve.jpg (73.11 KiB) Viewed 121 times

Heater
Posts: 14240
Joined: Tue Jul 17, 2012 3:02 pm

Re: algorithms and ethics at the Barbershop

Sun Jan 19, 2020 1:25 pm

Awesome. And both British. Hurrah !
Memory in C++ is a leaky abstraction .

User avatar
rpdom
Posts: 15882
Joined: Sun May 06, 2012 5:17 am
Location: Chelmsford, Essex, UK

Re: algorithms and ethics at the Barbershop

Sun Jan 19, 2020 1:58 pm

Heater wrote:
Sun Jan 19, 2020 1:25 pm
Awesome. And both British. Hurrah !
Yes, but not made at either of our local factories - Chelmer Valves or EEV unfortunately. I did once try for a job at EEV to work with military night vision units which needed a 6KV supply from a battery pack. During my interview they showed me how they were made and I tried one out in a blackout room. Fascinating stuff.

hippy
Posts: 6500
Joined: Fri Sep 09, 2011 10:34 pm
Location: UK

Re: algorithms and ethics at the Barbershop

Sun Jan 19, 2020 4:01 pm

Gavinmc42 wrote:
Sun Jan 19, 2020 1:32 am
In Law the Jury system relies on the combined life experiences of the Jury.
You could say those life experiences are the equivalent of human behaviour algorithm training.
But it still works under certain rules, innocent till proven guilty, no conviction just based on one persons word....
The beyond reasonable doubt is a rule for a black and white answer.
Law, and judging compliance with it, is probably one area where algorithms might be foreseen as potentially taking over bringing with it all the fears of algorithms if they do, including all the dystopian visions of AI determining if one may be predicted to be inclined towards law breaking.

We are already seeing a shift towards automated judgements with ANPR and speed cameras. That technology is far from true AI but some people are already unhappy, don't like the idea of black and white justice where there is no chance to convince a fellow human that a transgression was not severe enough to merit the punishment which would otherwise be given.

I think for most people the imminent fear is that AI - actually pattern matching - will flag them up through facial recognition as a shoplifter or similar when they aren't, and shops won't allow them in. "Computer says no" is what it comes down to, with no way of convincing it otherwise.

Many peoples have a belief that "rules are for the guidance of wise men and the obedience of fools" and black and white judgement doesn't sit well with that.

User avatar
rpdom
Posts: 15882
Joined: Sun May 06, 2012 5:17 am
Location: Chelmsford, Essex, UK

Re: algorithms and ethics at the Barbershop

Sun Jan 19, 2020 4:25 pm

hippy wrote:
Sun Jan 19, 2020 4:01 pm
Gavinmc42 wrote:
Sun Jan 19, 2020 1:32 am
In Law the Jury system relies on the combined life experiences of the Jury.
You could say those life experiences are the equivalent of human behaviour algorithm training.
But it still works under certain rules, innocent till proven guilty, no conviction just based on one persons word....
The beyond reasonable doubt is a rule for a black and white answer.
Law, and judging compliance with it, is probably one area where algorithms might be foreseen as potentially taking over bringing with it all the fears of algorithms if they do, including all the dystopian visions of AI determining if one may be predicted to be inclined towards law breaking.

We are already seeing a shift towards automated judgements with ANPR and speed cameras. That technology is far from true AI but some people are already unhappy, don't like the idea of black and white justice where there is no chance to convince a fellow human that a transgression was not severe enough to merit the punishment which would otherwise be given.

I think for most people the imminent fear is that AI - actually pattern matching - will flag them up through facial recognition as a shoplifter or similar when they aren't, and shops won't allow them in. "Computer says no" is what it comes down to, with no way of convincing it otherwise.

Many peoples have a belief that "rules are for the guidance of wise men and the obedience of fools" and black and white judgement doesn't sit well with that.
Jury service is something that I do have first hand experience of. Two years ago I was on a Jury for a potential murder trial. After much deliberation we eventually decided on a verdict of manslaughter. It was hard and took two weeks to get to that point. Later I found that the accused had previous history of assault, but obviously we weren't allowed to know that when we were making the decision.

Return to “Off topic discussion”