Robots and AI

Some people, interested in the future, can't stop thinking about the development of robots and general artificial intelligence. But I think most of them got it wrong, and that it is not going to happen that way.

There will never be human robots, except for marginal needs. If a robot needs to be completely general it would be designed as a human. But the idea of ​​making general purpose robot is wrong. People do not think about it, but everyday life is already full of specialized robots. I have several: one that cleans the dish, one who washes clothes, one that makes coffee for me, etc.

Do we really need general purpose robots? I don't think so. The day it will be possible to make "real" robots, we have already integrated our world with the virtual, and you can't distinguish between them anyway.

I believe it will be the same with general artificial intelligence. Some are obsessed with the idea of ​​making a general AI, smart as a human, that will then create the next, even smarter. One problem with this concept is to define "intelligence". There are already a lot of specialized AIs. Why should we make one that mimics a human mind? And if we create one, do we want it to be human? That means we have to give it the basic human goals, which are a will to live and a will to breed. Without those, it will not have emotions, and so it will not at all be human.

I think it will go the same way as for robots. The day we succeed, then we have already succeeded in making a copy of a physical brain in a computer. The need will be marginal.

So what about the Technological Singularity then? I don't think it depends on the presence of human-like robots or human-like AI. The Singularity is about accelerating progress, which is perfectly plausible with specialized AI.

The future of online gaming

As a generalization, there are two types of games; single player and online multiplayer. As more and more players are connected to the Internet, the demand for online multiplayer games will grow. When looking at online multiplayer games, there are two types; PvE and PvP. PvE means Player versus Environment. This is a game you play in cooperative mode with other players. PvP means Player versus Player.

There are some fundamental differences between PvE and PvP. In PvE games, huge resources have to be invested in by the game designers to create an environment for exploration and adventure. The average cost for a video game 2010 was more than 20 M US$, and typically lasts for 60 hours.

PvP games, on the other hand, can engage players for far longer time, even if the game design is much simpler. It may still take a big effort to produce the game, but there is less need for complex environments. On the extreme side, there are Chess, Go and other classical board games.

How come players return again and again, to the same PvP battle ground? The answer is the opponents. You don't like games that repeat the same every time. You want variation and increased challenges. In PvP, you can expect the opponents to improve, posing new challenges. That is why you can spend your life playing only a game like Go, and still learn and appreciate new ideas.

There is one market that has, to a large extent, been ignored. That is PvE games where players can add to the environment. I played the PvE side of World of Warcraft for several years, and enjoyed most of the time a lot. But in the end, you had to wait for Blizzard to create the next expansion. Imagine if it would have been possible for players to add content to this world, for everyone else to enjoy. The game could have been unlimited. World of Warcraft isn't easily adapted for such a strategy, though.

There are games with player created content. Many game engines allow for mods. One of the most successful multiplayer game, Counterstrike, started out as a mod to Halflife. If you look at what is happening with strategy game mods, like Starcraft, you will find an extreme ingenuity. Players have a way to use features in the most unexpected ways to create intriguing games and challenges. The problem is that these types of mods are of a temporary type. You play the game, it finishes, and the next time you start from the beginning again.

I predict a future type of game with the following characteristics, and think this is going to be the biggest genre in 10 years.

1. It is online, massive multiplayer.
2. You login, progress, save the state, and log out. Next time you return, you keep what you had.
3. Players allocate a territory in the game world and add content to it that other players can explore. That way, the basic game is for ever developing, and it will never be the same from day to day.
4. The adventures you create are your own, and no one else can modify them (unless you cooperate with friends).
5. The social aspect is a key as you will do the exploration together with online friends. Friends can be a set of people you choose to associate with, or it may be temporary groups you meet inside the game.
6. The owners of the game have to create a game server engine, a client, a starting area, and a list of flexible construction blocks.

In the 90s, I created LPMUD (with a lot of support from friends). It was a text adventure fulfilling many of these characteristics. LPMUD did not support graphics, and so it was difficult to move it into more modern settings. A contemporary version of this game type should support thousands of simultaneous players, which requires huge resources. For the first time, hardware and communication are now powerful enough.

This is the background to Ephenation.

The success of Chromebook

What are the chances for a success of the Google Chromebook?

There are 5 main segments today: Workstation, Laptop, Internet tablet, smart phones and mobiles phones. The Chromebook will have to compete with these, and possible create a new segment. But I think 6 segments is at least one too many, and one of them will be marginalized. Unless the Chromebook do not out-compete either the Laptop or the Internet tablet, I think it may be that the Chrombook is the one that will be marginalized. (Aside from that, I think workstations and "dumb" mobile phones are also going to be marginalized soon.)

An important question is if the Chromebook is right in time. For it to succeed, people have to accept that everything is done in the cloud. The cloud services are improving quickly, but is it quickly enough? It is a paradigm shift, and it may take time to get used to it. I think this paradigm shift will win eventually, but the success of Chromebook is more uncertain.

Why is Google investing so much into Chromebook? Google's revenue comes mainly from Internet advertisements. They prefer people to use Internet services for their daily businesses, instead of legacy applications downloaded to the PC. The perfect world would be if everyone only use a web browser. And so, a perfect world for Google would be if everyone use a Chromebook.

People are not going to change to Chromebooks just because that is what is best for Google, especially if the pricing is not competitive. I think Google will have subsidize the Chromebook to succeed. An alternative solution to a Chromebook would be to have a separate keyboard that can be attached to an Internet tablet. That looks more attractive to me, personally.

Simulation argument

There is a theory that we are all living in a computer simulation. Maybe not computer, but simulation anyway. This is a theory that is neither provable or disprovable, so it is undecidable. That means it falls into the same category as a religion. Nevertheless, there are signs that suggests we are in a simulation. These are based on what I would assume how such a simulator would be constructed.

Problem: To make a simulator of a real world, a lot of performance will be needed. One way to achieve this is to use a massively parallel computer. One problem with such a simulation would be that every event that happens anywhere would influence every other part of the simulation (through one of the forces in nature). This would be expensive, and has to be minimized.

Solution: Limit the speed with which events can influence other parts.

Our world: No information can travel faster than the speed of light.

Problem: Given time, events will eventually have effects on all other parts of the simulation, even if there is a speed limit.

Solution: Let the simulated world expand in such a way that there is a limit where changes can no longer catch up outside of a limited sphere.

Our world: The universe is expanding, and we will never be able to see beyond a certain point because that point is moving away too fast.

Problem: Computing exact results with infinite precision takes infinite time.

Solution: Impose a limitation on the precision of computations.

Our world: There is something called the Heisenberg Uncertainty principle. It states that precise inequalities that constrain certain pairs of physical properties, such as measuring the present position while determining future momentum; both cannot be simultaneously done to arbitrarily high precision.

Problem: There are many computations that may never be observed and may never have any effects outside a limited environment.

Solution: Make use of a lazy algorithm. Things are not computed until the result is actually needed somewhere.

Our world: In quantum physics, there is something called quantum entanglement, whereby the degrees of freedom that make up the system are linked in such a way that the quantum state of any of them cannot be adequately described independently even if the individual degrees of freedom belong to differentobjects and are spatially separated.

Exponential growth

Can the economy continue to grow exponentially?

There are many organizations today that claim that the growth must halt soon, or we will deplete important earth resources. Looking at it that way, it is obvious that, for example, the oil consumption can not continue to grow. And there are many other resources that simply can't be expanded on.

However, the thing is more complicated, and it is wrong to generalize. So let's have a look at some exponential behaviour. One of the more famous examples is Moore's law, which says that the number of transistors on a chip growths exponentially. A more interesting version of this law is the measurement of the number of CPU operations you get per dollar. This doubles approximately every year, and has been doing so for a long while. Actually, the trend is picking up speed. One reason for this trend may be that producers of processors knows where they will have to be in the future if they will be able to stay in the business. So every generation provides a measurable improvement compared to the previous generation, which means there is an exponential growth. This growth may not depend in depleting earth resources, at least not the way that energy from oil depends on oil resources.

Now on to the next type of exponential growth, and that is inventions in general. Sometimes, an invention builds on another. The invention wouldn't be possible until now. This is also a kind of exponential growth, as improvements build on other improvements. It can be argued, and it has been many times in history, that we now know most of what can be invented, and so this type of exponential growth will halt. Not because of lack of natural resources, but because of the end of scientific progress. This is harder to prove or disprove, but I am optimistic and believe we are no way near the end.

Finally, there is emerging another type of exponential growth because of inventions, and that is software. There is a constant stream of software inventions, where some of them build on the previous ones. This development is also fuelled by the exponential growth of CPU power. I am convinced this is where we will see an explosion of activities in the future. Until now, computers and software has just been seen as one component in our daily lives, in the list of cars, houses, phones, etc. But I think the software business will expand relatively to the other industrial areas. Economists that only look at GDP and other similar key figures will miss it out entirely.

Will a general artificial intelligence be benevolent?

It is hard to predict what AIs will do, especially as we have problems to define what an AI is. If you take a chess program, most would define it as narrow AI. Good at chess, but worthless at other things. But how then do you make a general AI? And if you succeed to create one, what makes it do anyting at all? Why wouldn't it just sit there, thinking of something, and remain incommunicative?

I think one of the important building blocks of an AI is the driving force. The problem with chess programs is that the driving force is extremly narrow. It is something like "for each piece, try every possible move. For each of the tests, try every possible response. All the time, evaluate the situation according to a strictly defined algorithm". And some more. This doesn't give the program much ability to generalize.

If the computer instead would have a driving force more generally defined, as "play human opponents, goal is to win", then it would leave more room for out-of-the-box solutions. For example, you win more if the humans play worse. So one solution would be to select low talented opponents. Another solution that would fulfill the goal would be to somehow arrange to have the opponents killed, and win by walk-over. A program that can take this kind of very general definition of driving force doesn't exist yet. Even if the given example is extreme, more near future realistic examples can be imagined.

It is a version of the old adage "the way you ask something determines what kind of answer you get". I think the defined driving force, the algorithms, and the rules, are what will ultimately make up an AI. You don't really need the rules to make an AI, but I strongly encourage the use of them. Even if the driving force is seemingly benevolent, it is no guarantee of the eventual behaviour.

A discussion you see now and then is if the driving force should be copied from humans. Looking closer to this, I think the main human driving force is to spread your genes. Everything else is just a side effect, to improve the chance that you succeed. Behaving nice to people, even doing charity, improves the chance.

The hope is that using a human driving force for an AI would also give it the same behaviour. But I think that is a mistake, as the nice part of (most) human behaviour is only a side effect. There are spectacular examples of humans that succeed well on an individual point of view, but catastrophically from the point of view of the human civilization.

Intelligent spam

In the near future, maybe 5-10 years, we will start to see intelligent spam. There will be participators in discussions argumenting for a position. The discussion will be two-way, not the one-way we see from spam today. This intelligent spam will be created automatically by software, from goal and rules defined by a stake holder.