Sure you can make an argument that it is efficient and you can take in more information, but it seem to me that that would dilute thought, analysis, and what it means to be a human.

Having a software package analyze and provide context would probably push people further into pigeon holes of thought. The software analyzes what you like/believe and it in turn echos that back to you endlessly until all you see, hear, and read are things that reinforce your own unbending view of the world. Sure the AI could grow and evolve, but only if the person continues to do so.

It’s a difficult problem, granted. However this relates to my point about the interface. In giving an algorithm a user teaches over time enough information about the context, you should be able to, using the learned algorithms, probabilistically choose what information to include in rotation with existing information. In the image’s case you may give two contexts because the probabilities are so close. Every day scene recognition software gets better the more training data it has and can handle. Numenta or like.com are an excellent example of this.

Humans infer context of other humans everyday based upon past experience and emotional cues we see in others. We may not get that right every time, but it’s a valuable social tool that aids in relationships and learning. It’s only through time, inference and ranking do we learn who are the experts and who are those to avoid (and you may be wanting to avoid my thinking right now: it gets worse . I see no reason we can’t digitally infer based upon enough of these cues, at first in narrow contexts, to suggest information. Sites already do this poorly today with recommendation engines limited to a very narrow subset of knowledge about us.

Short-term with the interfaces and algorithms we have now I don’t believe a great deal can be inferred to benefit us, but long-term I do see much of this happening.

Eventually I envisage sensor devices like eye wear that sees what you see, hears what you hear with see-through optical projection to display additional information for any given context. Some people call this mixed-reality, and there are already rudimentary examples of this. All this has to be managed by the users themselves however as to what information is really relevant in context, with the algorithm improving as they interact. The system learning, suggesting and personalising in a never ending cycle.

Whether it’s a computer with webcam, GPS, etc or head wear that can monitor blood flow of your brain and watch your context, it’s just a matter of how much can be inferred and how well that inference knowledge is then managed.

These days some new cars even have proximity echo sensors to tell exactly where you are in the car, and the first MIT quantum radar experiment has been performed with the potential to practically image through anything (your brain?): It’s all data that can be used to build an individuals context in order to infer what might happen next and suggest possible outcomes.

The problem is getting hold of all this valuable context information and then making use of it effectively. Sensors need to be meshed, data needs to be aggregated and persisted, and if this is to be context-based, then real-time processing needs to occur with resources traveling with the user. Second Life, for example, does this when you move from one island to another, moving avatars from server to server.

A lot needs to be done before any of this will ever have any noticeable impact though, but I do believe it will happen, one day, progressively.

Details here: http://www.vrmhub.net/vrm-hub-monthly-evenings/

I wonder how VRM might apply to politics as well as commerce: so instead of politicians and parties pulling support, voters would engage using tech to identify their advocates…

Is jan meet an open meet or does one need an invite Would like to attend, lemme know…

Al

As I point out in my post, context can meaningfully be processed only by a human mind. Hence the graphic at the end of the post.

In doing so, in a sense I’m creating a valuation or reputation for a source in a particular context. With the ARM Rank valuation being how many additional ‘arms’ it adds, or how well it augments me.

I start delving into the world of narrow artifical intelligence here as well as probabilistic logic when it comes to finding the simplest solution to a problem (like your supermarket example). What might be the right answer for you probably won’t be for me. Part of the difficulty is in the algorithm knowing whether you like or dislike something and the interface to feed that knowledge to it. It’s only in having this and ones historical data that I believe we can even begin to think in these directions, so I’m trying not to until I have that foundation. Although; it can be fun to do so!

ps. Augmented Relationship Management is an alternative name I’ve toyed with and this post reminded me of GoodGuide.

And that might change everything.”

It certainly will – especially for folk who choose not to tell the truth, in such matters.

Some assertions may need to be verifiable. Before you ask, I haven’t yet figured out how such verification mechanisms might work…