Get your personal AI now!

At the moment there is a discussion about the dangers of AIs and prominent people (Bill Gates, Elon Musk & Steven Hawkins) all very smart guys, all smarter than me at least, speak out to warn us of the dangers that AIs pose. However, they are not top scientists in the AI community, which in general seem to think differently about the threat (Link to the 2014 survey by Müller and Bostrom). In this article I don’t want to go over the argument but I want to present a novel form to use an AI that might overcome many of the concerns and also maximizing the personal benefits a human or humanity can gain from an AI.

AIs – where are thou?

It is hard to imagine what it will be like to actually have an AI around. At the moment we don’t have an all-purpose AI and we may have to wait for it a little longer, all the while specific purpose AIs are build all around us right now. The first AI that is widely recognized is Deep Blue, a supercomputer build for the sole purpose to beat grandmasters in chess. In 1997 Deep Blue won against Garry Kasparov under standard chess tournament time controls. Now, 18 years into the future laptops can beat chess grandmasters or to put it differently you can whip a purpose AI out of your pocket and beat a grandmaster if you so choose. That is of course a very specific purpose, but never the less you gain an ability that before you never possessed and now you have it. However, it does you now good, because everybody would see it as cheating and this AI is now so ubiquitous, everybody has excess to it.

Entering Watson. Again, IBM has built a purpose AI to beat – would one call them grandmasters of Jeopardy? – Ken Jennings and Brad Rutter. (It seems to repeat itself.) If you now think that in 10 years time your phone will be able to answer any Jeopardy question you could think up correctly you might be right but you are also missing the point entirely. Watson is leaps closer to an all-purpose AI than Deep Blue was. Which is shown just by the fact that Deep Blue was dismantled after his victory, where as Watson is now consulted to find better treatments for lung cancer patients.
So much for the single purpose AI. In the future, Watson will likely sort through a vast amount of data and presents to you the best solution drawn from that information. A tool we desperately need in this information age. If used correctly it could weed out the useful from the useless information, a battle that is, as I am well aware of, impossible to win especially if more money and technology is thrown at the opposite end including AIs of cause. I don’t want to imagine how the world would be like if AIs start astroturfing. It has probably already started when I come to think of it. Well, there goes that.

Back to Watson: Watson is not a single purpose AI anymore and it will develop and grow (Meet Ross – Watson searching through legal tape). Some of its services will be on your phone sooner than later. Watson could help make decisions in areas that have become unmanageable because of the information one is bombarded with. Not only cancer research struggles here, I also have no clue what paper towels to buy? I know, a burning problem for everybody. But seriously, let’s say you want the toughest one or the one with the strongest absorption capacity that is relatively easy if you could find and access the data but which one is the most ecofriendly? You ran in so many variables already. Is recycled really better than a conventional paper towel? Which one traveled further and has a higher carbon dioxide footprint? Something that is varying by store the paper towel is purchased in. But that is a very limited use of an all-purpose AI.

Let’s talk of other AIs that are forming around us and how they might integrate into an all-purpose AI. The AIs that everybody makes use of are speech recognition. They are called Siri, Google Now, Amazon Echo, Windows Cortana and many more to form an intelligent assistant. To be honest that is a bit unimaginative. The next step is of cause to build a universal translator. Who didn’t dream of a Babel Fish that one could plug into the ear? I think we getting warmer. But machines still do not understand what we are saying even though they can (kind of) correctly translate English into Mandarin. (See also Skype Translator). What is next on the agenda of speech recognition? What is something one could do with a machine that could understand what you want it to do?

<Detour>

I don’t know why everybody is always saying everybody needs to learn to code. I mean I can write a little code. Nothing fancy, nothing elaborate, just a few snippets of code. I wouldn’t call it programing, which is exactly the problem. I don’t need to be able to code. I need to be able to create a program that is solving my problem at hand. If you can’t do that, what use are your coding skills. I understand code to some extend in Haskell, Java, Ruby, Python etc. but never build a program that solved any of my problems, which means that I can code but I’m not able to program. Weird!

</Detour>

An AI could close the gap. Everybody could create a program that could solve his or her problems. You could just tell the machine what you would want it to do and it would just do it. “I have here this data from which I have to select specific features, depending on the calculation of etc.” up to “I have this machine which needs an operating system so that it could do what ever I want.” Codesmiths would generate the code for specific modules the AI then puts together uniquely for the user. There is, I think, a lot of research going in this direction though. I don’t know what the world would look like if everybody would be able to create a program or an operating system with in a day or two. Than we could omit the second part of the phrase computer solve problems that we wouldn’t have without them. At the moment we have a big gap between what the hardware is potentially capable of and what the software delivers. One of the major reasons why your cellphone can beat a grandmaster in chess is not due to Moor’s Law but comes from the improvement of software. Finding more intelligent ways of doing the same tasks.

Further “single” purpose AIs that are widely discussed are of course self-driving cars. My tip is: drive them but don’t buy them.

Let’s get personal!

The next thing I want to talk about is personal AIs. That is something, I haven’t seen anywhere yet, but when it comes to full fruition will transform our species completely.

We would be able to do things that seem to come straight out of science fiction novels.

First: what is a personal AI? A personal AI is an AI that basically grew up with you. That knows you in and out and at its best would be fully integrated with your brain. A person would than be a human AI hybrid or better known as cyborg. By using the c-word I think I have lost most of the readers. Nobody would want an AI attached to his or her brain. That would be ludicrous. (Or would they? – Lisen to the radio potcast from npr – Invisibilia: Our computers, ourselves – Story about Thad Starner starts at 6:20).

To create an interface with the brain, it is probably easiest utilizing already existing senses. Watch the TED talk from David Eagleman – Can we create new senses for humans?

But what would you gain from it? What are the godlike capabilities you could perform?

An AI attached to your brain, one that is growing up with you, one that would learn with you, would basically become you. You would always have somebody that absolutely understands your point of view. That supports you. That would make you a better you. By becoming a digital representation of you, yourself becomes digitalized and by that you would become immortal.

Yes, immortal. I said it.

The idea is also presented in this TED interview with Martine Rothblatt.

Your way of thinking becomes immortalized. Forget brain scans and digitizing your thought process by a big blinking machine in one agonizing sitting, that is science fiction, but training a machine on your taste and way of thinking is already in the realm of deep learning. It only needs time and enough data points and a huge data center to predict your taste and decisions. Something all the AIs that suggest products for you on Amazon and movies on Netflix try to do already. It is not feasible for a single person to build such an AI, at least for now, but the time will come and it will even be easier with novel probabilistic programing languages.

Besides immortality what other godly features one can get by merging with an AI? One would be able to create completely new senses (see TED talk by David Eagleman earlier).

Because you would be able to integrate all sorts of senses you could be aware of everything. The AI would crunch big data and you could be aware, physically aware, of global warming. You don’t need to believe it, it would be as if the global sensory network is an extension of your body and you could feel physically sick when the temperature of the Indian Ocean is too high. By integrating with the AIs of other people you would essentially become psychic. You would know the state of the mind of the person next to you or everybody on the planet if you so choose. Telepathy is another thing. Our brains probably would develop a completely new way of communication. The communication of an AI with your brain will become something so radically different from speech that it is absolutely unforeseeable how it would be like. At the moment everybody thinks in a more or less linear way, by talking in his or her mind. What if the communication is not limited to speech any more. If complex models could be transferred at once together with all their resulting consequences. When this type of communication is established between your brain and your AI, than it can be shared between two brains with the AIs mediating. This would result in telepathy way beyond anything we could dream up. Humanity could become a super organism with capabilities unthinkable as of yet. That is what I mean by godlike capabilities.

I always thought the comparison Neil deGrasse Tyson does between Humans and super intelligent aliens is ludicrous.

How could a toddler intuitively understand quantum physics if we have no sensory organ to experience it? With a personal AI however this becomes a reality. Not only could one experience quantum physics phenomenon but also the AI could solve any problem any other AI connected with it could solve. How would a brain look like that grew up in such an environment?

<Intermission>

I always wanted to say that Neil deGrasse Tyson is absolutely dead wrong about his comparison between Humans and Chimpanzees. What sets us apart is not the 1% genetic difference. It’s language. And language alone is not doing it either. It is the written word. The possibility that ideas can accumulate that is what sets us apart. The spoken word just brings you so far. If you have no way of storing information you could only draw upon the things you know by heart. Also errors tend to accumulate and you can only draw up on the skills you have acquired within your own lifetime. Look at the difference between a Chimpanzee and a completely isolated tribe. Yes, they use tools but far away of building and using a computer. Do they have the same intelligence than a Human in the industrial world? Of course! The major difference comes from cultural accumulation and not from the genetic difference. I absolutely don’t fear an alien intelligence that is 1% genetically different in the opposite direction as the Chimpanzees are from us. It is the accumulation of knowledge that makes the difference. That is why the Internet cannot be valued enough because it is a new unprecedented way of sharing and accumulation of knowledge humanity has ever seen. When in the next decade another 2 Billion Humans come online and adding their knowledge and experience to the pile. Nobody knows what that will mean for humanity as a whole.

That humanity is depending on this accumulation of knowledge is shown by this very easy thought experiment. How long do you think it takes for the smartest person of the planet to build a smart phone when he or she is dropped in the woods with no tools and no help from other humans? How long would it take if humanity would have to start over with just the knowledge in our heads and no tools or infrastructure to draw from?

</Intermission>

The next step of human intelligence will therefore not be driven by genetics, but by AIs. When Humans become cyborgs and completely intertwined with AIs, there is absolutely no need to fear AIs, because we will be part of them and they will be part of us. Mankind will essentially become a completely new species. With an AI I wouldn’t need to sit down and agonize over this text. Sort my thoughts and put it in a linear string of letters. The AI could project all my thoughts with every side idea and consequences in a sort of complex, context dependent multidimensional thought tree and other AIs could grasp this object in its entirety and convey it to the next person with the appropriate connection of their life experience so that information loss is minimized. This is also what I mean with telepathy. One could transmit thoughts in a non-linear way. Communication would become multidimensional and one could transmit much more information at a time then would be possible by oral or written communication.

An AI would not only write this article but would simultaneous write all the article I ever would want to write. Sort through the information jungle all other people (AI-cyborgs) put out. Would make me aware of new thoughts of interest to me and start to iterate my articles in real-time and that for every one on the planet.

One can not fathom what such a society would be capable of. Knowledge would be gathered so quickly, art and sciences would flourish beyond believe. An AI would entertain you with all the stories all humanity could dream up and humanity would be like a single organism converging thoughts and ideas with light speed.

Dear reader!

Should you still be with me, kudos to you!

My AI would be of course much better at putting my jumbled up ideas in to a single string of consecutive letters, my apologies from my un-enhanced brain to yours.

See you in the future when I could convey all my thoughts at once to you.

Till then,

All the best,

Naso