In this article I will describe a new way of understanding Artificial Intelligence 2.0 as a social platform where simple global human interactions allow the emergence of greater forms of Intelligence. I will explain the basic principles of what I consider the foundations of AI 2.0 and finally, I will present a concrete example of what a real social expert system would be like.
It has been long mentioned that Web 2.0 services “harness collective intelligence” but what’s most interesting is the higher forms of intelligence that emerge when people interact with these services:
|Web 2.0 service||Direct benefit for user||Emergent Intelligence|
|http://del.icio.us||Save personal bookmarks using tag classification||
|http://last.fm||Listen to personal radio stations||
|http://youtube.com||Upload and watch videos online||
Web 2.0 has brought us a great collection of principles and patterns (tags, wikis, social ranking systems, etc) that allow us to build richer cloud-based services that extract the whole potential of crowds. Wikipedia, Facebook, MySpace, and YouTube are just the first generation of Web 2.0 applications that we can build with these new patterns but the possibilities of combining them are endless.
Networks that look too much like neural networks
Web 2.0 social networks have grown a lot in past 3 years. Currently there’s over 60 million users registered in Facebook and 200 million users on MySpace. Millions of people around the world are getting virtually interconnected. Millions of links between people are created and destroyed every day. Links get stronger and weaker (through favorite lists, comments, rankings, etc) among the virtual entities. Huge and complex networks have been created out of the simple of interactions of people.
It is inevitable to compare the structure and behavior of social networks with our biological neural networks and not realize the similarities that they both share:
- They are both made of a huge amount of relatively “simple” interconnected entities
- They both evolve over time without any central point of control
- They both connect their entities with “dynamically weighted” links
- They both have entities with a “LIFE” of their own
- Higher levels of “intelligence” emerge out of the “simple” interaction of their entities
What is Artificial Intelligence 2.0?
In traditional AI (1.0) systems, a human employs a computer to solve a problem: a human provides a formalized problem description to a computer, and receives an (intelligent) solution to interpret. In AI 2.0, the roles are often reversed: the system collaborates with a large number of people to solve a problem, then collects, interprets, ranks and integrates their solutions.
One of highest expressions of Web 2.0 services are Social Networks. These networks basically interconnect people, but, what if instead of interconnecting people we interconnected (live) ideas? Ideas that link to other ideas with “dynamically weighted” links. Ideas that have a life of their own.
An example: Social Troubleshooting
Whenever we face any type of issue that requires “troubleshooting”, we mainly work with 3 different types of independent knowledge:
- SYMPTOMS: Signs of a possible error in our system.
- CAUSES: Possible root cause(s) for a SYMPTOM(S).
- ACTIONS: Steps that solve a specific CAUSE of an error or just extra troubleshooting steps that will generate more SYMPTOMS.
In Social Troubleshooting we would use 3 independent and linked databases: SYMPTOMS, CAUSES and ACTIONS (instead of just one huge unstructured text-based database as we store knowledge nowadays). Items from the SYMPTOMS database will be linked to one or more CAUSES, which, at the same time, will be linked to one or more ACTIONS, which may generate even more SYMPTOMS. The more links we have among these items, the greater the weight will be for these links, just like we see in neural networks. Users can choose among many possible paths to solve their issue but the system:
Principles of Artificial Intelligence 2.0
AI 2.0 is based on 4 core principles:
#1 – Social classification
“The only group that can cateorize everything is everybody“
Through the use of tagging users should be able to classify content and obtain many benefits from doing so:
- Tagging associates keywords or terms with content in an attempt to describe or classify it for an increase of usability.
- Creating a solid taxonomy can define a structure that helps users find what they need and allows them to easily classify new content, instead of fitting content in a fixed hierarchical classification.
- Creates a more flexible structure that allows the classification to adapt to the content and to be scaled more easily.
- When users classify content they are actually also classify themselves.
- Massive amounts of intelligence can be extracted from tag-based systems.
- More info: http://en.wikipedia.org/wiki/Tags
#2 – Social ranking
Users know what is best when they see it. Social ranking allows AI2.0 to boost to the top the good ideas and bury of bad ones. This can be done via a simple voting mechanism. Each idea can be voted up or down.
#3 – Social linking
This is one the most revolutionary component of AI2.0. In traditional Semantic Web or Web 3.0 systems pieces of knowledge were linked but the social component was completely ignored. If we take into consideration the frequency of the social linking between pieces of knowledge we would immediately obtain a neural system capable to learning and evolving in real-time with no supervision. The possibilities are endless and can be applied to any environment that could take benefit of social collaboration to solve problems: science, medicine, software troubleshooting, government, etc.
#4 – Social reputation
One of the lessons learned from the Web 2.0 is that users should be recognized if you want to keep their engagement in the system. Social reputation has demonstrated to be a huge incentive for individual contributors and it is a core component to keep the quality of the content and links. A good example is http://www.stackoverflow.com
What is the benefit for users?
- Receiving a fair share of the result
- Direct compensation
- Desire to diversify their activity (e.g. “people aren’t asked in their daily lives to be creative”)
- Esthetic satisfaction
- Curiosity, desire to test if it works
- Volunteerism, desire to support a cause of the project
- Reciprocity, exchange, mutual help
- Competitive spirit of a game
- Desire to communicate and share knowledge
- Desire to share a user innovation to see if someone else can improve on it
- Desire to game the system and influence the final result
“This is more than open source, social networking, so called crowdsourcing, smart mobs, crowd wisdom, or other ideas that touch upon the subject. Rather, we are talking about deep changes in the structure and modus operandi of the corporation and our economy, based on new competitive principles such as openness, peering, sharing and acting globally.”
WIKINOMICS, How mass Collaboration Changes Everything
…from Don Tapscott