As we have discussed with @deborah and @bengoertzel earlier, ability to withstand flow of misleading online information is existentially critical fo an intelligent being, be it human or robot or any artificial creature.
Risks are enormous, as everyone becomes target of massive “psyops” warfare field: https://www.cbinsights.com/research/future-of-information-warfare/
Big companies are trying to provide their own “centralised” solutions implementing sort of crowdsourced censorship: https://www.telegraph.co.uk/technology/2018/04/30/facebooks-plan-combat-fake-news-shrink/
However, having any sort of “thought control” be centralised and closed-sourced, makes it impossible to ensure it serves the community, not individuals. The “who will guard the guards” question arise.
The problem can be solved with implementation of kind of “Reputation Agency”, which would use public open-sourced crowdsourcing data processed by AI algorithms to infer reliability of particular sources of information and news:
https://www.youtube.com/watch?v=88Wzt5aelZ4
There are few approaches to solve the problem:
A) “integration of the reputation agency into widely used social media platforms” - just impossible;
B) create personal browser-kind agent filtering and annotating browseable content - huge but doable;
C) use standalone agent, accumulating content being posted, shared, commented and liked and use it for inferences - that is exactly what we are doing with Aigents right now, although access to data is restricted to extent depending on platform:
https://www.youtube.com/watch?v=NfS3LTcFm1k
Try what we are doing and provide your feedback!
I am wondering if there is any overlap here or possible synergies with the other two oracle or prediction market blockchain projects, Gnosis and Augur? Any thoughts, @akolonin?
Everything in AI and blockchain is synergetic and overlapping 
Seriously, the closest overlaps that I see are:
- ability of reputation engine to infer reliability/reputation of particular oracles/predictors/communities/predictive models as well as potential predictability or particular subjects of prediction
- potential ability of predictive product contribute to assessment of reliability of particular news events, so you kind of use predictive product to predict the present and the past because truth of present and the past is still vague
Hi @akolonin. I still need to observe the information on your website more thoroughly and I will definitely do it a bit later but I guess that I have catched the overall idea and I like this vision. The only thing that raise a question for me - have you ever considered the possibility of your website UI changing? I have no doubts that your system would be rather interesting and sophisticated under the hood but I think that nowadays it’s very important for customers how any website or app which they use looks like. Though I suppose that maybe sometimes it can be annoying for creative and enthusiastic person to think about the look of the software when there is so much to think about the concept and develop the realization itself but still UI questions are too important for modern audience. First impression plays a big role.
Absolutely. The user interface (running on demo web site at https://aigents.com/) is very tiny layer of open-source code which can be replaced/rebranded with the same functionality working under the hood. The core code is delivered as either server-side solution (https://steemit.com/ai/@aigents/aigents-power-server-for-personal-social-analytics-back-end) or standalone aigents distributions for Android and any desktop (with platform-specific UI).
Thanks for the answer @akolonin. It’s great that this interface is only some interim solution for demonstration as I understand. Because the idea itself makes good impression. I will try the functional and provide the feedback later. It’s comfortable that I know Russian so it would be easy for me to observe all materials on your Youtube channel. From a first look there is a lot of interesting information there.
There are some more bi-lingual materials organised here http://aigents.com/en/references.html
And more is on https://steemit.com/@aigents
In particular, for SinglularityNET, we are exploring idea of dedicated Reputation Agency:
https://medium.com/@aigents/reputation-system-design-for-singularitynet-8b5b61e8ed0e which can be part of SingularityNET framework or separate entity serving different online communities and applicable cases.
Note that current https://aigents.com/ functional does not cover the case in full. It is especially limited with latest Facebook ban on using its API for personal analytics, so the best experience of this product aspect needs use of Steemit: https://steemit.com/ai/@aigents/aigents-finding-opinion-leaders-on-steemit
Thanks for the detailed information 
Comment on first link: Most of this is your typical weakness for a big-data approach. Ben mentioned something about the tragedy of people encrypting their data, but if you build your AI right, it should still be able to operate without containing the worlds information.
It should be somewhat hardened from malicious reprogramming, while maintaining a Free and Open Source model.
Note, I do think some data collection is useful, but what if that data suddenly goes away? I like the chatbot approach personally, because it can continue operate with a limited amount of data.