Ghosts in the machine learning pipeline will be impossible to exorcise Tech rarely touches the soul. It enrages when social media pours petrol on hotheads. It inspires when Hubble comes back to life and delivers more cosmic awe. It pays our bills when we work in it, and it empties our pockets when we drunk-eBay that vintage console game collection. But when it brings back a dead lover? That's no …
It seems we've come a long way from Dr. Sbaitso.
A new kind of immortality is now within reach, and if the history of our culture is anything to go by, we will find the temptation irresistible. .... Rupert Goodwins
Be fully prepared to accept that irresistible temptation has already comprehensively successfully captured the hearts and minds and provided all the necessary tools and vital virile ingredients for sublime and supremely engaging applications of the phenomenon ...... and against which there is no known possibility of real physical defence or practical virtual attack.
Successfully comprehensively captured the hearts and minds of whom and/or what is the gazillion dollar money shot question one always will find is safest best left securely unanswered ..... for some knowledge is deadly dangerous to know and share unwisely.
"amanfromMars" always writes nearly meaningful gibberish. With this he steals processing time from your irreplaceable life, as you foolishly try to find sense in his output.
I suspect that some of the other people I meet in chat contexts had a similar motive, and died some time ago leaving their respective automatic provocation programs running.
Agree. Real dictators would not be happy to be "virtual", that won't satisfy their egos. And knowing that they are not in control wouldn't help -- any power outage will cut them from their loving subjects.
Also being virtual won't let them grab any pu.... never mind.
"Bezos was this month reported to be a significant investor in Altos Labs, an age-reversal firm which is on the scientific quest for immortality."
https://www.theguardian.com/commentisfree/2021/sep/10/jeff-bezos-eternal-life-amazon-taxes
Rupert,
Thank you for a thoughtful and thought provoking piece on a subject that had slipped under my mental radar.
Uses of AI to recreate a person who we have lost is as you say an almost irresistible temptation and like many things, like fire, can be two-edged.
Both good and bad will come of this and I feel that not only the AI community should get a grip on the potential problems but the whole of society. One way would be through legislation but I have doubts if politicians are properly equipped to deal with it. Maybe some sort of commission or inquiry would help, that way we could gather the thoughts of everyone. All of us will be affected by this and it is only right that everyone is consulted and it is not left to vested interests to twist any controls on the technique to their private advantage.
Speaking for myself I would love to have been in a position to re-connect, however ephemerally, with people I have lost and I can see that handled in the right way and with sensitivity may well prove to be of great help to many people
Yeah, there are some grim ethical dimensions to this that make me feel that the realized version of this idea may not be very good for people.
That said, when faced by the sudden and irrevocable loss of a loved one, I could see how a person could find a measure of solace from an avatar of the person helping them to move on. That last point is pretty important. Most of the portrayals of this idea point out the many ways it goes wrong though.
I fear that we are more likely to suffer things like the chatbot in the last season of Westworld, with models architected to manipulate, and to string their user along into dependency. Worse, very little in our legal system would prevent the operator of one of these systems from trying to convince the target that the person was still alive.
Indeed, you can't.
You should not, however, create a fiction based on the data of the dead.
I understand that people have trouble letting go. I understand that very well. That's the reason all those spiritism charlatans make their money.
But you have to face the fact : your loved one is dead. It's harsh, it's unforgiving, but it's a fact.
Entertaining yourself with a fac-simile is nothing but soul porn.
Deal with it. Your life is finite, just as their's was. Live your life, don't live in the past.
Hmm.
Assuming I have the time, I not infrequently spend hours every week driving imaginary cars around a facsimile world, whilst pretending I can win races. I find it quite relaxing, especially the Australia one because I used to live there, and because there isn't one set in NZ, which would be better.
If I was, for example, to spend the same time talking to a facsimile of a lost relative, why would that be worse? Both are, objectively, a total waste of time that could be better spent, er, reading stories (most likely about imaginary people in variously rendered facsimiles of the world or some derivative thereof), watching TV dramas or films (ditto). Or I might spend hours a week at worship in a building dedicated to a most-likely imaginary supreme being (or beings) of some kind. I could apparently even buy some dubious looking "dating sim" games; as Steam seems to want me to "discover". I might spend hours listening to made up stuff in poems, or music lyrics.
All fake, all virtual, all imaginary.
Humans tend to spend a lot of time doing virtual things, and almost all of them are really not the worse for it. So quite why the possibility of a virtual grandad (or whatever) gets everyone in a froth is beyond me.
Sure, just like with many things, those few who get lost in the imaginary will probably need some help. But this isn't a new or very interesting problem; it's just people losing their way. And, I hazard to guess, being dependent on talking to a simulated friend is way less toxic than (e.g.) gambling addicts who think they just need - or will get - just one more decent "win", or for serious drug/alcohol users who think it seems to help them cope.
I'm don't think I would want a simulated relative, but I'm not going to tell other people what they should want.
Although perhaps it might be amusing to generate a simulated me, so that I can imagine haunting the internet after I'm gone. I'd have to train it to pretend to drive facsimile cars in an imaginary antipodes, though.:-)
Re this title; So far it has not been done, only because a sufficiently powerful and wealthy entity hasn't yet tried.
I can easily imagine a number of egos who are currently and allegedly alive wishing to continue their influence from beyond the veil.
Attaining some kind of legal recognition of their 'personality' in a program would be the beginning, if the program could pass a Turing test, someone, somewhere will try to get it the legal status comparable to a corporation.
Can you imagine Feacbook being run by a Zuckerbot for eternity?
There, is one of the biggest threats to the world from AI, giving it ghost with legal rights.
I was bereaved earlier this year; my wife died very suddenly and unexpectedly.
I have ample material from online chats conducted while we were courting and engaged (we lived on opposite sides of the world, so all our conversations were electronic). Her Facebook profile would provide ample material to bring it up to date. In other words, I am in a position to do exactly what Joseph did, given the skills (which I don't have).
However, I am not even slightly tempted. It was my wife's intelligence and sometimes quirky response to things that made her what she was. It was doing things with her, sharing common interests, that made our life together. It was helping her cook her favourite Chinese dishes - the list goes on and on.
An AI avatar might perhaps fool me for a few minutes, but even if it got everything right, it still wouldn't fool me for long. How does an AI handle the idiosyncratic English of a person for whom it isn't their first language? What about all the private things that never came anywhere near an electronic interface?
All I'd end up with would be a sort of Barbie-doll version of my late wife. Better to accept that she is gone, move on and see what lies around the corner - as she would have wanted me to.
Paul, agree with your assessment. ITIRC it was GK Chesterton who said that "a man looks to someone addressing him hoping to hear the unexpected" or words to that effect. That is exactly what you pointed out. Otherwise one indeed has a mere replay or worse, another twatter chatroom. Condolences on your loss
To ensure that if any of my relatives decide to engage in this kind of behaviour, they don't get to inherit.
Ideally, I'd like a legally binding way to prevent anyone using my digital legacy (though my footprint is as small as I can reasonably make it - the majority of my online posting is here!).
While appreciating the feelings of the bereaved, I would not want them to think that anything cobbled together from my previous ranting and raving in any way represents my opinion on, well, anything - particularly after my death.
Of course your relatives already have an internal model of you, probably including any ranting and raving you might have done in their presence (or reported to them in conversation), in their heads already. As does, to a very small degree, any attentive reader of your thoughtful and intelligent posts here.
So you've already lost the battle against idiosyncratic, dubiously constructed, and partial renderings of yourself, as have we all. What's one more?
OpenAI knows this, which is why they and anyone else with half a sense of foreboding are so hot on developing AI ethics"
AI ethics are at best a misnomer. The best that can be done is to implement an inevitably rule based system based on the ethics of the AI programmer and/or trainer (which is of course an arbitrary set of ethics, not one spontaneously developed by the machine as humans do).
In the conclusions to In AI We Trust: Ethics, Artificial Intelligence, and Reliability published last year, author Mark Ryan states "One can rely on another based on dependable habits, but placing a trust in someone requires they act out of goodwill towards the trustor. This is the main reason why human-made objects, such as AI, can be reliable, but not trustworthy, according to the affective account.", making the important distinction between reliability and trustworthiness.
It turns out that there's a cluster of human talents that are inherently missing from anything that ultimately runs on a Turing machine.
The "AI" output described here is based on the conversation scripts of the two parties, and they own the copyright on them. For the lady who died, the copyright passes on to another person in the same way as all her other assets. This is all well established law.
These scripts are essentially the source code for the program, and the AI thingy is essentially the compiler that turns it into computer-readable code.
"Expect at some time in the future to start receiving advertising and/or calls from a 'friend'..."
I've been getting those for years ... Scam telephone calls claiming to be a friend or relative using the same generic phrasing that so-called "fortune tellers" use. And yes, these days they are usually bots.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2024
