People understandably have a lot of questions about how exactly artificial intelligence works. How could a machine understand nuance and context like a human brain can? Is creating Artificial Intelligence (AI) that sophisticated really possible?
Your answers lie with AI systems like Watson: IBM’s boundary-pushing technology, a collaborative effort designed to amplify human intelligence. Watson is on the cutting edge of cognitive computing, which means it thinks and learns.
So how, exactly, does it do that?
Before you can teach a machine to process information like a human brain, you need to understand how a human brain thinks. People go through four steps before making a decision. They can happen almost instantly, but we do it every time.
1. Observe: We take in information that we find around us. Whether it’s visual, audio, or written, we take in everything around us.
2. Interpret: We analyse the information we’re gathering and make sense of it. We create hypotheses about what is happening in front of us.
3. Evaluate: We go through those hypotheses and find out which are right or wrong to learn more about the situation.
4. Decide: We know exactly what is happening, and choose how to react.
This is how we learn how to operate in real life. Through trial and error, we find the best routes to take.
Watson is designed to go through that exact same process whenever it’s presented with a problem to solve. This is how it learns so organically. Through these steps, Watson is capable of understanding, learning, reasoning, and interacting with the world around it.
It’s almost as if Watson went to school and earned every business degree there is. It’s capable of understanding almost any industry, and has even been specifically trained for certain industries — everything from education to financial services to healthcare.
AI technology, unlike traditional programming, is capable of interpreting and reasoning like humans do. It uses context to recognise the differences between a bat that is used to hit a ball, and one that flies through the air for example. This allows us to communicate with Watson in natural human language, rather than having to talk like a computer.
Because Watson learns over time, the more it focuses on a single area of expertise, the more useful it can be. And because its services are hosted on a cloud platform, it can be integrated into almost every touch point of an organisation. It’s digital, so it goes wherever you do.
This ability for AI to think, learn, interpret and reason is why many organisations are using systems like Watson as virtual agents to liaise with both customers and employees, and help them solve problems or get answers to questions much faster.
The way IBM have trained Watson is as impressive as the technology itself.
Watson is trained by real, live human beings. This allows it to learn not just general knowledge, but also the deep-cut things about the areas it needs to be an expert in. The human teachers give Watson a very large amount of data that is clearly labeled. Then these experts go through the information with Watson and refine the output, helping it to make sense of what it’s learning.
By going through many cycles of input, output, and preening, Watson can start to make sense of what it’s seeing on its own.
Yet many fear the training of AI systems are open for bias based on the information being fed and the individuals training the systems. IBM believes that AI actually provides the means for bias to be mitigated or even engineered out of systems. When AI systems demonstrate inherent biases, they are reflecting the data that people give them. They are confirming – and making visible – bias that already exists. Researchers at IBM believe such bias can be algorithmically detected and corrected and are working with organisations to ensure their AI strategies seek to mitigate bias. This area of research is absolutely critical to trust and transparency of AI.
This is why Watson, and AI in general, can never replace people, as some have questioned as the tech has developed. Watson works best as a tool for people, and learns better through the collaboration of man and machine. For example, Watson learned how to understand casual dialects and interpret tone with Natural Language Understanding and Tone Analyser, respectively, which is useful in education settings when a language translator is needed. There’s also a Watson Virtual Assistant that can act as a chatbot for businesses, which can aid in things like customer support and even banking.
Possibly most important is Watson Knowledge Studio. This program allows people to train Watson in specific industries or areas of expertise on their own. This program is easy to use and will really make Watson the sidekick your business needs.
Source: https://google.com / https://mashable.com