Personal preference swarms and the quantification of intention

http://darkai.org/?p=243

2 Likes

I hope this gives some additional clarity on how DAVS could work. Intentions / preferences could be broadcasted anonymously. Every virtual citizen has wants and needs. Because it can be anonymous no one has to know in particular what all your wants or needs are or who wants or needs what in order to supply it.

Also there would be no need to shop. The process of trade could be automated entirely by automated agents acting on behalf of virtual citizens. This would remove a lot of the exploitation in the form of subliminal ads or other tactics while empowering all virtual citizens.

Every virtual citizen would have a personal preference bot net. This bot net of theirs would be their “swarm”. Each bot would be able to interact with all sorts of different open protocols from blockchains to email to decentralized value exchange networks such as Ripple.

These bots would be able to exchange secure communications. So for example Amon’s bot could communicate with Dirk’s bot over any protocol where their bots meet each other. The message would then be relayed until it reaches the person with the private key.

Trades could take place as well and anywhere. A bot to bot trade could be conducted on this forum for example and none of us would know who owns either of the bots. This would be possible because the bots in theory should be able to do transactions over many different blockchains or protocols.

Thoughts?

I like that, but this comes with a wider spectrum of possibilities I hope, because your example of markets are limited to direct buy, you haven’t mention a case where new offers are made.
Perhaps it would need a separate tool to reference products (or other things), but seems interesting.
Now, I would like an example over something else than buy/sell, does it has one?

1 Like

Every Bot should be able to trade/swap anything for anything. So that would mean it can swap different tokens. So if you have Bitcash then you could give it to your personal preference botnet but if you have a lot of vouchers, giftcards, etc, you can delegate to your preference preference botnet to get them to trade in some profitable way.

Bots would be able to do any sort of market transaction that you can do so I don’t see how they’d be limited. Trade mean all sorts of stuff such as auctions between swarms of bots, reverse auctions, job auctions where bots bid on jobs to allow for a decentralized job search on behalf of individuals.

There are many possibilities here which is the whole point.

It exchanges information, if this might consider “free for free” trades of information then limits to information are surely considered to avoid overload.

Note aside, it would be an improvement over torrents if an implementation over what to catch is added to a project like this, like a reference list of potential transactions where to select what to offer in exchange of seeding something. And could help marketing if done well.

Now, I have not clear if this offers a clear referencing list of transactions, but again this can be done in a separate tool, but I wouldn’t wait too much to create one.

1 Like

I think the general idea of this topic is quite fascinating. It reminds me of what I have proposed within the Doctrine Zero mailing list quite a while ago about what I called “interactive markets”. Interactive markets are based on a close cooperation between consumers and producers, rather than the passive reactive behaviour of consumers in the “reactive markets” we have today. Intent casting is the first step needed for interactive markets. It would be a very useful tool even if no AI agents were used!

Now, when you include AI agents you can indeed bypass the need for “shopping” once the technology is advanced enough. The other question is whether consumers actually want that, because shopping is a rewarding experience for many. But I assume there is a veritable subset of consumers who want to automate at least some part of their “shopping”. In that case, I see a problem with completely automated exchange between bots: Humans might not be really happy with the deals the bots come up with, so you might want to include an option to make them confirm a deal proposed by the bots first. Or add the possibility to cancel deals within a certain period of time after they have been made by the bots.

Apart from the technical considerations, I propose to avoid the use of the term “bot net”, because that it usually associated with a cloud of machines infiltrated by hackers and used for malicious purposes. I assume what you mean with that term is a network of AI agents working for a single person, or a group of persons. A term like “AI helpers” (instead of “bot net”) might sound much friendlier and more descriptive.

Just signatures, like hashes. The Internet of Things? Example http://telehash.org/

Or you could just use the blockchain itself. Information would have to be transferred such as an sms message, email message, post on a forum with a QR code or some meta data. But all the trade information would be in the meta data including a description of what is being traded.

Circular Multilateral Barter and Offer Agents

https://web.archive.org/web/20140212065551/http://www.supplydemand.info/blog/the-concept/

http://p2pfoundation.net/Circular_Multilateral_Barter

We could call them robot delegates instead but I can see why bot net might have some negative connotation. The problem is if we don’t call it a bot net then no one in the world will know what it is and less people will use it.

At least people know what a bot net is.

The idea is to solve the problem of attention scarcity. Some humans who have a lot of time and money do like to shop. Those humans can still spend their time shopping.

But the vast majority of humans don’t enjoy shopping because they aren’t rich so for a lot of humans it’s just about survival, acquiring resources.

Conditional preference networks on their own might not be perfect. Learning conditional preference networks on the other hand will learn what you like, what you want (wishlist), what you need (survival needs) and it can find the best deals on your behalf. You can set it up so people can have as much or as little automation as they want, as much or as little attention given to their bots as they want, but I think these personal bot nets are going to be in our future.

Also the problem with shopping is most people aren’t rational. Marketing exploits our limited attention spans and our inability to decipher complicated labeling. Most people don’t even know what they are buying, most people don’t have time to read the list of ingredients for everything they buy, and most people don’t know how everything they buy was made.

Bots could factor in all of this. Connect the dots: https://www.youtube.com/watch?v=wE2XEw9wY6I

1 Like

http://darkai.org/?p=251

That Provenance DApp looks really useful and promising. Looks like this whole interactive market scheme could actually be implemented within the next years to some degree.

Maybe it would be good to think about what components this system would need and who could work on each of those components.

I still think we need a better name for “personal preference bot nets”. What about “Personal AI brokers”?

How about “personal swarms”?

I really like the term “personal swarms”. It feels empowering to command a swarm of AI agents that do useful stuff for you. And it’s something of a paradigm shift, I think. So far, we have been thinking in terms of singular AI agents, or leveraging the swarm intelligence of humans. But if we can get AI agents that use some kind of insect hive intelligence, we get something really new and powerful. However, is there already something like a proof of concept for personal AI swarms?

There are many different swarm intelligence simulations which can show how an AI swarm can work. For example particle swarm optimization, ant colony optimization, flower pollination optimization, all which are good for different purposes.

For example flower pollination optimization works very well as an “attractor” for decentralized apps (DApps) and DACs. Crowd funding for example can utilize swarm intelligence patterns inherently.

If you agree with me that this is the right direction to go then maybe you can help research swarm intelligence applied to the matter of decentralized applications?

1 Like

That’s fascinating stuff. So, in terms of algorithms there seems to be a significant basis out there (we might also consider how slime molds find the shortest path through a labyrinth). But is there actually some useful software out there that already applies these algorithms to real worlds problems, or is it still all on a very theoretical level?

In any case, if we want this to take off, we need to get to the point where the average user finds it easy to create a swarm of personal AI agents for specific purposes. There’s certainly a big untapped market potential (if it could be all open source, it would be even better) for that.

And yeah, the topic of how swarm intelligence and AI can be applied to decentralized applications sounds like a very fresh and interesting topic, but I have other priorities right now. It sounds like something which computer science majors and enthusiastic hobbyist programmers would love to research and write about, though. So, we need to tap pools of such persons. If I had some real spare money I might want to make a competition for writing an essay about that topic. The results might be quite interesting. But there are some ways to get good thinkers to write about such topics even when no money is involved. The recent Transpolitica ebook proves that point. I think with your connections you might be able to trigger some people to write blog posts about this fascinating topic, Darklight.

1 Like

Radivis here is my latest blog post.

http://darkai.org/?p=267

Nice. So, it’s “personal preference swarms” now :slight_smile: Now let’s talk about more technical issues: Do we have any interested people who could program such swarms or is it all theoretical right now? Or is the plan to forward the idea to groups of people with the necessary skills to implement this idea?

I’ll try to program some of it depending on what can be done with Ethereum. Right now I don’t know yet how to go about it.

Maybe part of the DAVS prototype, maybe not.