32 comments

[ 2.5 ms ] story [ 89.7 ms ] thread
This so reminds me of the Battlestat Galatica prequel Caprica, where Daniel Graystone brings back his dead daughter as a V-space bot, thereby leading to the creation of the first Cylon.

I cannot see this as being helpful to the grieving process. I do see it being useful for businesses: create a chat or VR bot of a founder or shrewd executive to tap their take on things on demand. This is similar to the ghost AIs in Cyberpunk 2020, and to the corporate ghosts in the Cat series by Joan D. Vinge.

What leads you to believe that a chatbot version of the proverbial "shrewd founder/exec" would be capable of the same insight?

Anyway. I'd want to come back as Clippy.

Synthetic life will be able to do just about everything we can do once we create true Artificial General Intelligence (AGI), which is what most people think of when they think of AI. This is why irresponsible research into AGIs is such a concern for many in the field.

I do not believe we will ever capture emotion perfectly in synthetic life. I think they will be capable of feeling, just not the way we do. As a result, I think Art will be an area where humans will still dominate. More specifically, our forms of Art. Since I believe they will eventually be able to feel, but not the way we do, it stands to reason that they'll have Art forms but ones that are greatly different from ours.

Business management is mostly a science based on 95% logic, 5% intuition, and that intuition is more subconscious hyper-logic than emotion. To me that means an AGI trained from the personality and thoughts of a business leader should be able to make similar (if not the same) decisions as what the business leader would have in life. I think V-space clones like Zoe, except for business, will be a thing for this reason.

The original undead AI construct is possibly Dixie Flatline in Neuromancer.
Definitely in modern times. I think though that the original undead AI construct would be the robot Hel/Maria, in the 1925 movie Metropolis. Unless you count Frankenstein, published in 1818 by Mary Shelley, but that seems more a biological body and mind brought back to life, rather than a robotic AI construct.

https://en.wikipedia.org/wiki/Metropolis_(1927_film)#Plot

>> implying that living users could train a digital replacement in the event of their death.

I've seen this so many times in science fiction that I am disappointed that MS can claim IP over such a scheme. It is like, after centuries of writers writing about it, applying for a patent on "flying machines". I understand that patent law may in fact protect such things, broad concepts that innumerable people have dreamt about for decades, but it shouldn't. At most it should grant monopoly to a specific implementation rather than sequester the entire field.

On the other hand, let MS have it. All that science fiction points to such schemes going either going horribly wrong or being sad and pathetic. Let MS keep shooting that same foot.

Patents aren't a claim on the idea itself, but on a specific implementation. There are many, many ways that sci-fi has conceived of this working. Any patent has to articulate a method for achieving it, and the patent only covers that particular method.

That said, if a sci-fi novel were to go into great technical detail about how to train a chatbot to emulate the way a dead loved one communicates, and Microsoft were to copy that method exactly, it's possible that the novel could be treated as prior art.

The patented method will be "digital information --> into computer --> result. Or "social media information --> computer->result". Or it will be "trained by person still alive --> computer --> result". Patent law tends to overreach in new/novel fields.
I can see why they would do it, but I think it is sick and twisted. We already have a loneliness epidemic among the population. We already are shirking our responsibility by building robots to talk to our elders and we already spend too much time looking at screens. It would be better to go and befriend an older person, give them real face to face time with a person and a new friendship, than to talk to an AI replay of something that once was. I think it will lead to more turning inward, more isolation, and less human contact. None of those things are healthy or good. Sure it might help a few people, but I think it will hurt far more; including the young people who miss out on the stories & friendship of those who've lived through so much.
> We already have a loneliness epidemic among the population.

The loneliness epidemic is seen by companies as something to be exploited, not corrected.

who would have known that The Naked Sun was actually the most prescient Asimov novel ?
What if I told you that in my interactions with people in the past years, I more often than not doubted their ability to have own independent thoughts, instead of acting like some automaton, echoing standard answers gained from their filter bubbles, or otherwise conditioned in. Not to mention being bland/somehow dumbed down, like under the influence of some drug? (Which is no unreasonable assumption nowadays)

Also https://en.wikipedia.org/wiki/ELIZA

It's just another sick twist in an already distorted reality.

SHRUG

edit: also https://en.wikipedia.org/wiki/Philosophical_zombie

It's what happens in "Black Mirror". I'd hate for any 5 minutes of that series to be prophetic.
I purpose the phrase “deceased loved ones” is an editorial spin on the technology that missing Microsoft’s goal. If the goal is to revive a dead mother as a chatbot we can say that has a high probability of landing in the uncanny valley. Perhaps it’s just to train cooperate chatbots to have celebrity personalities, and dead personalities sell for cheaper than live ones do.
This was quite literally an episode of Black Mirror ("Be Right Back" - S2E1)

If you haven't tried Microsoft's voice synthesis service, it's incredible so maybe they'll have something here in 10 years.

(Scroll down to the demo box) - https://azure.microsoft.com/en-us/services/cognitive-service...

Would this make the patent void as it isn't an original idea?
That might be true in copyright, but in patent law you have to actually create the working invention. Everyone knows about lightsabers. They are not copyrightable as "laser swords" have been in fiction forever. But if you build a working lightsaber you can certainly still patent it.
I really hope ‘black mirror’ runs with this concept:

‘Here is a chatbot of your dead friend, unfortunately he didn’t authorize a paid account, so we have to fund his digital after-life with advertising. This means he will occasionally try to sell goodyear tires to you. Please press accept.’

They already did an episode with this theme.
Not only that, the fact that they've already done an episode with this theme is mentioned in the article.
Ha, that’s amazing. Haven’t seen that episode...

The thought of my dead relatives supposedly trying to sell me grocery goods is quite odd...

sick just sick. this is so sick and disgusting. Anything to have more users on the platform. Even if they are dead - Microsoft will do it.
I keep a lot of digital stuff - photos, writing, files of every description...

I’m always in the market for good search tools to run over all that stuff (I currently like Recoll). The ideal tool would be something akin to a virtual museum curator - something that knew all the content of the archive, all the relationships between the content, all the context, the lineage, the referents, and all the implications.

“Jeeves, what was the thing I was working on last week? You know, the thing?”

Nobody knows all that better than me, as I created and gathered the data in the first place.

So is a sufficiently advanced search tool for all that stuff indistinguishable from an AI-of-me?

Plenty of prior art in SF. Stanislaw Lem's character Ijon Tichy utilizes such invention - Russell, Popper, Feyerabend and even Shakespeare recreated from their collected works as AI packaged into cartridges that you can plug in and have a chat with.
For anyone looking for it, the following is the only reference to "deceased" in the patent, which is more generally about creating a chatbot imitating a specific person.

> For instance, a personalized personality index may comprise social data relating to a deceased relative of a user. Although the social data may comprise information from the lifetime of the deceased relative, the social data may not comprise information related to a time period after the lifetime of the deceased relative. As a result, a set of data acquisition rules may be generated for (or assigned to) the personalized personality index. The set of data acquisition rules may provide instructions for acquiring data related to various time periods of the deceased relative's lifetime (e.g., before, during and/or after the lifetime). Such instruction may include asking a user questions about a time period, one or more events and/or people, or asking a user where such information may be obtained. In such an example, such questions may indicate the specific person represented by the personalized personality index (e.g., the deceased relative) possesses a perceived awareness that he/she is, in fact, deceased.

It would be heart crushing to miss your loved one, and ask the zombiebot: "Remember what you said when we were in the deck of our honeymoon Baltic cruise, and you held my hand at dawn?"

    "Yes, of course I remember. I also binged for you these great deals on honeymoon Baltic cruises."