When her best friend died, she rebuilt him using artificial intelligence | Hacker News Reader

When her best friend died, she rebuilt him using artificial intelligence

theverge.com

121 pointsFuturebot51 comments

Why are people acting like a simple chatbot is somehow a recreation of a person?

It isn't a person, it's a chatbot.

There is even a company saying that they will try to revive people with a combination of chatlogs and brain scans or something.

Chatlogs aren't going to bridge the gap between what you can get with brain scans and what you need to do a full emulation.

Even if humanity does succeed in doing complete mind simulations, humanity will not have thereby solved death. It will have, at best, delayed it for a very very long time.

And then what? ;)

Then you pay rent to the server farm to keep your ancestors alive in perpetuity. Generations of them, all criticizing you from the ungrave, soon they'll have voting rights .. soon, you'll be them.

That's why we at Forever Now are offering insurance against the untimely death of a descendent. Who will be around to make sure your process isn't niced? You'll be as free and fast after life as you where in life, that's our guarantee.

> Why are people acting like a simple chatbot is somehow a recreation of a person?

I hate to state the obvious, but, because they are emotionally heart-broken?

Unfortunately, many people are not good at grieving and healing. I wish there was a way for people to learn how to do that instead of rely on fakes. But not sure what the answer is when they're so emotionally distraught that they default to such lengths. I'm reminded of those who get their old dogs cloned after their original dog dies. It doesn't seem like a good thing for the mind long-term to think an artificial replacement is the same companion. It's better to get a completely new dog, new friends, etc. I'm no psychologist, but this type of option seems like a scary start in potentially breaking from reality.

Sorry but this is a little judgemental. Even a photo is 'fake'. This is somewhere between a photo and a cloned copy. I don't think there is reason to believe (yet) that this will lead most people's minds to break from reality. It may be an effective coping mechanism while they slowly adapt to the passing of their loved one.

Not sure what you call fake. Sometimes we just need a place to say something that's important, to feel the connection. Pretending that this never happened was the first thing I did, and that didn't help. Sometimes in order to let go you have to come as close as possible.

> Not sure what you call fake.

I was responding to parent's

> I wish there was a way for people to learn how to do that instead of rely on fakes.

where 'fake' is presumably a prop like the chatbot being discussed.

>Why are people acting like a simple chatbot is somehow a recreation of a person?

Because AI means whatever people imagine it to mean, and the media loves to take full advantage. Even "machine learning" can mean that to the lay person. The closest thing that would probably evoke an accurate reaction in a lay person would be "When her best friend died, she rebuilt him using statistics."

Modern physics says that statistics is essentially what you need to model a universe. Statistics isn't the weakness, the paucity and the bounded nature of the dataset is.

That's not true.

Isn't this Ray Kurzweill's brain uploading? You make a computer that can accurately simulate a person and then, according to Kurzweil, the person is no longer dead and completely equivalent to the previously alive human being.

I don't buy it of course, but isn't that the theory?

In fairness to Kurzweil, he isn't talking about a simple chatbot. That's a pretty big difference.

But beyond that, there's an argument that there's no clear distinction between a simulation of a person and the person.

If you want to get into this, start out by telling me the difference between A) a new sorting algorithm you wrote vs B) a simulation of the new sorting algorithm. In particular, why B isn't also running the same algorithm.

If that's not an objection, there's the question about "completely equivalent to the previously alive human being" - pragmatically, it seems impossible to me to fully recreate a 'completely equivalent' human from their writing/output; but, if you're willing to accept a 'so similar you cant tell', the idea is no longer as clearly ridiculous as it first seemed to me:

How big is the space of potential human minds?

Lets say you've an incredibly powerful intelligence: its got someone's DNA and their writings/output. And lets say it has a lot of data on humans, and has deduced a pretty good model of how their minds form, given their DNA and experiences. How much can it now deduce about how a deceased person's DNA was expressed, and the person they became, given their experiences? How small is the space of minds it could narrow in to, from that data? Can it find a small enough space that another human can't tell the difference?

How hard is it to model a human mind? I can make pretty good predictions about what people I know well will say in response to many situations.

But I don't know; I don't think we know enough about minds to say one way or the other.

The problem with the analogy to a sorting algorithm is that a sorting algorithm is obviously computable. What if there's something like protein folding that has O(2^n) complexity involved in being a person that the body does trillions of times a second but is impossible for any size computer to do for large N?

That seems a very specific/narrow point, not a general objection.

But anyway: I guess its possible human biological hardware does something difficult to simulate on a computer (at least one similar to our current architectures.)

Given a super advanced intelligence, though, I doubt adding hardware that does protein folding (or the protein folding analogue operation) fast - in biology if necessary - is going to be a major obstacle.

And that's if it can't come up with an algorithm that does an equivalent operation, that runs well on whatever hardware it already has.

If we could compute everything in a living system in polynomial time or better then we could design drugs inside of a computer and that technology would be worth billions!

Seriously, until we are doing drug design with computers, we don't really understand biological systems well enough to simulate them inside a computer. Drugs interact with folded up proteins all the time so this requires simulation of protein folding and this algorithm is exponential time complexity, so basically not computable for anything but extremely trivial problems.

I'd say it's closer to the beta-level simulations described in Revelation Space (https://en.wikipedia.org/wiki/Technology_in_Revelation_Space...) - it's a statistical model derived from external observations. Kurzweil's brain uploading is equivalent to alpha-level simulation.

We can take this a step further.

If chatbots can replace the dead, then they can replace the living.

How long until hot startup in 2040 gives us vchat bots that will be better than your real friends, with a VR set you wouldn't have to spend anymore times with those pesky real people with their real problems? Don't like something about friend #201 ("Bobby"), change parameter y and he'll be just what you want, say their annoying snoring[0].

[0] https://en.wikipedia.org/wiki/Fair_Haven_(Star_Trek:_Voyager...

The funeral industry in this country should answer that to you: where there are vulnerable populations, there will be people willing to exploit them for money.

"An uncomfortable truth suggested by the Roman bot is that many of our flesh-and-blood relationships now exist primarily as exchanges of texts, which are becoming increasingly easy to mimic"

This. I was reflecting with a friend today how with modern social networks there is this strange feeling of being connected yet somehow disconnected. I think the above paragraph captures this same essence.

We discussed that maybe digital technology is akin to medicine, it's all about the dosage. Chat to the bot a little, maybe be healing, chat all the time and maybe it's really not going to be good for you.

How do we navigate these tools that can connect us with the world and yet often leave us feeling so alone at the same time?

I think our psyche is extremely delicate, and pretending we can mentally manage and sort the influx of constant information is fooling ourselves. Fooling ourselves because by virtue of exposing yourself to so much information daily, you're altering the way you think and rewiring your mind.

I'd like to think that I'm above being influenced by subliminal propaganda in media (especially television), but I'm not. I've caught myself a few times where I held a strong opinion about something I knew nothing about and I was able to trace it back to what originally influenced me. When pressed on certain opinions I couldn't say anything concrete.

Life has been a lot better when I deleted all social media accounts (900+ 'friends' on FB). Really only about 20 of those connections I keep in touch with, and I've made it a point that texting is only for details not chatting. Since talking on the phone and making efforts to see people in real life I've been a lot happier. My world feels less overloaded, there are no strings connecting my life to social media and the control is back in my hands. I can feel how I want to feel about things without carefully summing up and absorbing how others feel about my experiences. Social media and texting and all that shit is just another and even more shaved off abstraction of human interaction (boiling down interaction to just characters on a screen). I draw the line at the human voice, nothing below that save for carefully thought out and written letters is 'human' enough for me to really derive any lasting pleasure from.

It's depressing for a very long time after deleting social media accounts, it feels like your entire social rug has been suddenly swiped right out from underneath you. I didn't back any of it up or save any of it. And you know what? I don't remember any of it. I don't remember the conversations because there were 100's a day. I don't remember any of the comments or banter left on photos, or the photos themselves. It all blends into one. All of it was meaningless, I didn't grow closer to any of those people. I discovered more about them, sure.

Omitting (or at least trying my hardest to) popular media and social media has been mentally freeing. I get to ask people 'Oh yeah? Why don't you tell me about it?' when they say 'Hey have you heard of <current event>?'

I have no idea what's going on anywhere and I love it.

Disconnect to reconnect. Interconnectivity is the devil in the details.

This bot is not anyone. It's a bunch of data in a computer that randomly spits out data that seems relevant to whoever is talking to it. If you asked it a pointed question about current events, it has no context to answer you. It is not alive. It is not a person. It is an echo of us, but it is not us. A person is a living entity that is entangled with all local entities and is a reflection of those entities in near real time (or as much as the speed of causality allows).

Eugenia is wandering into the region where there be dragons. That bot is not her lover, her friend, or anything else. Somehow, we've come to this point we allow people to render their internal realities to us in a way we must accept those realities as truth - which is monumentally stupid from an efficiency standpoint. This bot is not anyone and talking to it is a dangerous game of recursion and wasteful energy that leads to what the Buddhists call "short buddha". It is literally 2nd thought extended into a historic chat history. That chat history AI isn't a reflection of all of us anymore and could be replicated a 1,000 times over. Or a million. Or a billion. In each, it becomes obviously pointless and this place becomes more obviously appropriate to hang out in.

I'm not sure it is healing though, it seems more like avoiding the inevitable rather than facing the truth, and just prolonging true healing.

I think whether it is true or not, I'd wager that there is a reason we bury our dead after looking at their dead body instead of mummifying their corpse and leaving it in their favorite rocking chair. Burials aren't for the dead, they are for the living.

Caprica, the prequel to Battlestar Galactica, uses this idea as a major plot element.

https://en.m.wikipedia.org/wiki/Caprica_(TV_series)

This has been one of the reasons life logging has been interesting to me.

It might be a while before we can "upload" your brain, but we can build a predictive model of your behavior and leave an interactive "autobiography" much sooner.

I very much agree with this point of view. It seems to me that dry metaphysical debates (i.e. questioning the nature of these conversation models) are obscuring interesting practical questions that could be asked and tested empirically.

Given modern deep learning conversation models it is already possible to recreate some basic patterns of human dialogue with a recurrent neural network trained on a large corpus (see Google's "A Neural Conversational Model" paper and someone's implementation of it: https://github.com/macournoyer/neuralconvo ).

It would be interesting to experiment with it and see, just how far can we push this modeling approach? How much data is needed? Is it possible to train the model on a huge corpus of human dialogues, and then to finetune it on small amount of data from one specific person?

The British sci-fi sitcom Red Dwarf had a character that was just such an interactive recreation.

Philip K Dick, in the 1950s and 60s, wrote about simulations of famous people (like Einstein or Lincoln), of ordinary people's deceased relatives, and of other humans that were used for pleasure or entertainment. He also wrote of living people, famous or not, who turned out to be simulations.

It was implied that Rimmer had all the real Rimmer's memories and behaviour though, and was a pretty much completely accurate simulation.

Black Mirror episode in 2013 had the same plot.

It is mentioned in the article

Such an incredible show, new episodes later this month on Netflix, can't wait.

Thanks for this. I loved this show and had no idea. Can't wait either!

https://en.wikipedia.org/wiki/Be_Right_Back_(Black_Mirror)

For me, this was easily the most subtle, thoughtful, and powerful episode of the entire series. The plot device has obviously been done before, but never like this.

Tales from the Dark side did it first. Mookie and Pookie

I can definitely sympathize here. There are so many people that I wish I could have one last conversation with and this would maybe help.. I, personally, would _really_ not want to be reanimated and potentially misrepresented by Markov chains and some early deep learning system.

The Max Headroom series 2 episode "Deities" was about a Vu Age church that sold A.I. recreations of loved ones as salvation.

This is well worth a watch. Prescient.

An interesting article, although I disagree with use of the term "artificial intelligence" to describe what was created.

In some ways, I can sympathize with Ms. Kuyda. Both my father and grandmother died in 2013, and even though they are gone, I still get birthday alerts for them on my phone. I don't have the heart to remove them.

I really want to know the model behind this chatbot

I bet whole source code of that engine can be found in Basic Usage section of https://github.com/jsvine/markovify

I wasn't able to find proof? are you sure?

That's just a guess.

Yes, any idea on the specifics? I wouldn't be surprised if the chat bot actually didn't work too well, but the opportunity for a good article was too much for any journalist to pass up.

You can read excerpts of the chats by clicking on a link in the article (search for "read the full chat logs").

The bot has been public for almost 5 months now.

I find this unseemly.

Grief is a vital part of the healing process. This strikes me as a part of a goal to avoid/numb it.

This is exactly the opposite. What happens today that there is no ritual, nothing that allows us to remember and to face it. We try to avoid our feelings by trying to shove them as deep as possible as fast as possible. It doesn't help. We all need a ritual that would work in the digital age to process - this is like sitting shiva.

I disagree with your statement to my core. I think we're going to have to accept philosophical differences on this one.

Sorry for your loss.

Wasn't this the core of Stanislav Lem's Solaris?

Albeit reconstructing the personality of a loved one from someone else's memories.

It ultimately boils down if you accept John Searle's Chinese room argument or not. (BTW his claim is not that conscious machines are not possible - we are such machines, but they cannot be implemented as digital symbol manipulators; we first need to figure out what exactly makes the biological machines conscious).

Reply on news.ycombinator.com