Decentralized Artificial Intelligence
If we look today everything is centralized and limited to particular owner of that entity like Facebook, Amazon, Wikipedia, WhatsApp, YouTube etc. Like they solve problem or challenges using their own dataset/approach, they don’t actually look for generating common solution for societies problems. This had lead to emergence of centralized solutions, these websites come up with innovative ideas and become more powerful platforms because of services they provide. Like for example now a days WhatsApp has updated its privacy policy which tends to have all the personal data of user which is indirectly helping them in generating advertisements or some references of products by using textual analysis or some neural networks due to which users sees such advertisements/product references on their social media and which urge them to buy that product. So in this way only such companies have emerged as robust centralized platforms. They have also emerged as centralized server where in all the data come in and goes out.
So, for the magnification of the future of Artificial intelligence we have to focus more on collaboration. Here collaboration relies on finding domain, group of people who are willing to share their knowledge to solve a complex society problem but which leads to risk of centralized solution-oriented organizations.
Benefits of Centralized Artificial Intelligence:
Let’s see how these Centralized Artificial Intelligence systems exploit the huge datasets they gather for their benefit. As of now all AI solutions are centralized which involve a risk for companies that implement them, but some of the benefits they get from them are:
The Rich Get Richer:
· Many of the AI solutions in the market are provided by tech giants that control the huge datasets. These datasets keep increasing constantly as their dominance in public is more and hence more the data they accumulate, the more datasets they produce.Dataset Owners Influence Populations:
· Centralized AI systems can influence populations and direct knowledge based on their willing. Most common scenarios include asking for location, asking for personal information in order to get tailored and the best suggestions for them.
To overcome these benefits of only Centralized Artificial Intelligence we must focus on decentralized Artificial Intelligence
The Decentralized Artificial Intelligence System
Decentralized Artificial Intelligence is a way to bring innovative ideas, datasets based on intelligence to solve set of complex problems. This decentralized Intelligence model will provide opportunity to large entities / companies that control these huge datasets to be independent and will also expedite the upcoming affluential companies help make new innovations grow.
Let’s now see the comparison between Centralized and Decentralized Artificially Intelligent models:
In (Fig(a)) above, all AI model development work is controlled by central authority(X). Let us assume that Model A is used for clarifying images containing text and model B is character recognition tool which generates machine encoded text, while Model C translate this encoded text from one language to another and Model D is the actual text to speech tool which will process machine encoded text to particular language and read it out, but in this case if a noisy image is being passed to the centralized model then this model will not work properly and yield correct results and also the data required for this model we will have to be follow sequential steps from model to model for proper execution, whereas in decentralized model as show in Fig(b) all the model development is done independently and is not controlled by some entity where in all the models work independently of each other. So, if a noisy image is passed to Model D of Fig(b) and since the input cannot be process by Model D same as the centralized model, then in decentralized routine the Model D will request the Model B to process the noisy image and send the machine encoded text, but then what if the Model B finds the text not clear same scenario as in centralized model? In this scenario, Model B will then send it to respective Model available with the capability of handling noisy images better for clarification and similarly extract the machine encoded text and later send it to text-to-speech model. So here we do not have the requirement to follow sequential procedure, rather we can allow the new start-ups to help us better the solution.
As demonstrated by above example, we can get a glimpse that decentralized Artificially Intelligent model helps the bigger entities / companies to embed their models / datasets into single architecture and work with the upcoming affluential companies to find the best solution possible and solve various complex problems.
Conclusion / Findings
Ultimately, decentralized Artificial Intelligence will provide a hierarchy of achievement that is created through the advancement of knowledge. At the root of this achievement is individualized collaboration. An application of which is already being implemented known as Synapse AI. Synapse AI provides a machine learning marketplace and exchange. This marketplace allows developers to contribute data, train machine learning models, and be compensated for both. When disparate parts come together to share their findings, new or consistent understandings can be presented to help overcome challenges and, thus, unlocking decentralized AI’s true potential.Thus according to my opinion both the structured models are good but if we want to solve major complex problem quickly and easily we should work towards the decentralized artificial intelligence.
Contributors:
Shubham M. | Abrar M.
Aditya N. | Ayush R.
Muzamil R.
