Allow me, if you will, to take us years into the future (assuming humanity doesn't kill itself first). Humanity has been able to build androids.
Some are used for physical or dangerous work, others for service, other's still for 'personal use' (ew). They must be programmed to do their tasks.
What about 'care giving'.
To do so, humanity decides we have to program 'morality' into them, as morality is programmed into humans all the time - seems like a doable task.
For discussion:
Will humanity ever be able to program morality into a machine?
Why or why not?
Programming morality
Moderator: Moderators
-
- Prodigy
- Posts: 3187
- Joined: Wed Nov 11, 2020 11:32 am
- Has thanked: 1510 times
- Been thanked: 824 times
Re: Programming morality
Post #2Basically no, it isn't possible in my opinion. Machine's are algorithmic, digital computers follow - relentlessly and endlessly - formal rules, algorithms, they are glorified finite state machines.nobspeople wrote: ↑Mon Mar 21, 2022 9:46 am Allow me, if you will, to take us years into the future (assuming humanity doesn't kill itself first). Humanity has been able to build androids.
Some are used for physical or dangerous work, others for service, other's still for 'personal use' (ew). They must be programmed to do their tasks.
What about 'care giving'.
To do so, humanity decides we have to program 'morality' into them, as morality is programmed into humans all the time - seems like a doable task.
For discussion:
Will humanity ever be able to program morality into a machine?
Why or why not?
There are some interesting scientific and philosophical arguments that the human mind is capable of non-algorithmic processes, that is we are capable of solving what are termed non-computable problems, if this plays a role in morality then clearly a machine that is purely algorithmic can never solve these kinds of problems.
-
- Savant
- Posts: 9864
- Joined: Mon Feb 27, 2012 6:03 am
- Location: Planet Earth
- Has thanked: 189 times
- Been thanked: 266 times
Re: Programming morality
Post #3[Replying to Sherlock Holmes in post #2]
There is no reason why computer has to remain purely algorithmic, is there? It's as easy as adding in an analogue component into the decision path.
There is no reason why computer has to remain purely algorithmic, is there? It's as easy as adding in an analogue component into the decision path.
Re: Programming morality
Post #4Analog devices are algorithmic, they implement computable functions.Bust Nak wrote: ↑Mon Mar 28, 2022 4:30 am [Replying to Sherlock Holmes in post #2]
There is no reason why computer has to remain purely algorithmic, is there? It's as easy as adding in an analogue component into the decision path.
-
- Savant
- Posts: 9864
- Joined: Mon Feb 27, 2012 6:03 am
- Location: Planet Earth
- Has thanked: 189 times
- Been thanked: 266 times
Re: Programming morality
Post #5They can be, but they don't need to be. Your brain is an analogue computer.Sherlock Holmes wrote: ↑Mon Mar 28, 2022 9:59 am Analog devices are algorithmic, they implement computable functions.
- DrNoGods
- Prodigy
- Posts: 2716
- Joined: Wed Jan 11, 2017 2:18 pm
- Location: Nevada
- Has thanked: 593 times
- Been thanked: 1642 times
Re: Programming morality
Post #6[Replying to Sherlock Holmes in post #4]
How many SPICE models predict "exactly" how an analog circuit wil behave in the real world? Put a bunch of these kinds of analog circuits together and the ability to compute exactlly their behavior gets progressively less accurate. Component tolerances are one significant reason for this, and I can see how throwing in a bunch of analog circuits can lead to some level of unpredictability in the output ... the more complicated the circuitry and the more components, the worse this gets.
But they are far less predictable and with much more variation in outcome, expecially when combined with other analog circuits. And they are generally continuous rather than discrete in their behavior. I've designed many transimpedance preamps for optical detectors with a wide range of shunt resistances (and bought many from commercial sources), and they can oscillate when they technically shouldn't, saturate before they are "supposed" to, exhibit nonlinearities and offsets that differ from the datasheets by sometimes significant amounts, react differently to EMI sources, etc. And these are relatively simple analog circuits. Their behavior can also change over time as components age and drift in their R/C/L/beta, etc. values.Analog devices are algorithmic, they implement computable functions.
How many SPICE models predict "exactly" how an analog circuit wil behave in the real world? Put a bunch of these kinds of analog circuits together and the ability to compute exactlly their behavior gets progressively less accurate. Component tolerances are one significant reason for this, and I can see how throwing in a bunch of analog circuits can lead to some level of unpredictability in the output ... the more complicated the circuitry and the more components, the worse this gets.
In human affairs the sources of success are ever to be found in the fountains of quick resolve and swift stroke; and it seems to be a law, inflexible and inexorable, that he who will not risk cannot win.
John Paul Jones, 1779
The man who does not read has no advantage over the man who cannot read.
Mark Twain
John Paul Jones, 1779
The man who does not read has no advantage over the man who cannot read.
Mark Twain
Re: Programming morality
Post #7The brain does bear a resemblance to an analog computer but as to whether it "is" I don't know that I agree.Bust Nak wrote: ↑Mon Mar 28, 2022 10:44 amThey can be, but they don't need to be. Your brain is an analogue computer.Sherlock Holmes wrote: ↑Mon Mar 28, 2022 9:59 am Analog devices are algorithmic, they implement computable functions.
The fact is analog electronics devices always implement computable functions like summation, multiplication, differentiation and so on, given that our brains appear to sometimes be capable of evaluating non-computable functions, then whether it be digital or analog, we are still facing the limitation that these machines can only ever evaluate computable functions, this they can never achieve what our brains can.
-
- Savant
- Posts: 9864
- Joined: Mon Feb 27, 2012 6:03 am
- Location: Planet Earth
- Has thanked: 189 times
- Been thanked: 266 times
Re: Programming morality
Post #8Just like the neurons in your brain.Sherlock Holmes wrote: ↑Mon Mar 28, 2022 11:03 am The brain does bear a resemblance to an analog computer but as to whether it "is" I don't know that I agree.
The fact is analog electronics devices always implement computable functions like summation, multiplication, differentiation and so on
So can your everyday digital computer, it's just that sometimes they get them wrong, sometimes they don't halt. Our human brain faces the same limitation.given that our brains appear to sometimes be capable of evaluating non-computable functions...
Re: Programming morality
Post #9This is all true and although we can't predict their behavior we agree (I assume) that this is due to our mathematical model being overly simple or our knowledge of the parameters being less precise than we need, after all these devices must obey the laws of physics, that is everything is bound by laws.DrNoGods wrote: ↑Mon Mar 28, 2022 10:54 am [Replying to Sherlock Holmes in post #4]
But they are far less predictable and with much more variation in outcome, expecially when combined with other analog circuits. And they are generally continuous rather than discrete in their behavior. I've designed many transimpedance preamps for optical detectors with a wide range of shunt resistances (and bought many from commercial sources), and they can oscillate when they technically shouldn't, saturate before they are "supposed" to, exhibit nonlinearities and offsets that differ from the datasheets by sometimes significant amounts, react differently to EMI sources, etc. And these are relatively simple analog circuits. Their behavior can also change over time as components age and drift in their R/C/L/beta, etc. values.Analog devices are algorithmic, they implement computable functions.
How many SPICE models predict "exactly" how an analog circuit wil behave in the real world? Put a bunch of these kinds of analog circuits together and the ability to compute exactlly their behavior gets progressively less accurate. Component tolerances are one significant reason for this, and I can see how throwing in a bunch of analog circuits can lead to some level of unpredictability in the output ... the more complicated the circuitry and the more components, the worse this gets.
Morality implies free will to me, and free will seems incompatible with anything based on computable functions, even functions that contain randomness.
Re: Programming morality
Post #10That's the rub though, they cannot. Non-computable means there is no algorithm possible, not even in principle, no amount of skill and design can lead to a workable algorithm.Bust Nak wrote: ↑Mon Mar 28, 2022 11:10 amJust like the neurons in your brain.Sherlock Holmes wrote: ↑Mon Mar 28, 2022 11:03 am The brain does bear a resemblance to an analog computer but as to whether it "is" I don't know that I agree.
The fact is analog electronics devices always implement computable functions like summation, multiplication, differentiation and so on
So can your everyday digital computer, it's just that sometimes they get them wrong, sometimes they don't halt. Our human brain faces the same limitation.given that our brains appear to sometimes be capable of evaluating non-computable functions...
The simplest example of non computability is the famous "halting problem" this is where we are to devise an algorithm H that can examine some other algorithm X and tell us if X can ever stop, it is generally insoluble and there are many other more complex examples.
We cannot devise a set of rules that can tell us if some other set of rules always terminates or not, in other words H(X) is non computable.