Future of technology

Many might think the future is all about chatbots and voice-controlled tech. Theoretical physicist Stephen Hawking believes machines will one day become so intelligent that they will exceed their creators’ intelligence and become genuinely sentient. Read more What technology will we have in 2050?

The Conversation Africa’s Topline is a week-long series of events designed to surface and explore critical issues affecting Africa to a national and local audience. The Conversation Africa will curate topline stories in collaboration with the continent’s leading academics, policymakers, and scientists.

Future Tech

There are two kinds of AI, machine intelligence and neural network algorithms. Machines are capable of artificial intelligence, as are the computers we use daily to make our phones, car, and other devices work. The technology we use for this purpose has brains, meaning they’re intelligent. But this tech still only has AI in them.

Neural networks

Neural networks have layers. The layers models are after neurons in the brain that send messages through networks of neurons—the layers of a neural network model this communication so that artificial neurons can learn and evolve.

It calls “deep” because they have more than one hidden layer compared to the layers of AI we see worldwide, like Siri and Google Search. Deep learning is a form of artificial intelligence. It works by feeding more and more data into the network, and the network analyses and learns from that data. The network will gradually learn to improve its performance on that data to teach itself. It’s still in its infancy and is still a developing technology.

First developed Neural networks in the 1960s. However, because they were initially only used for processing data, they couldn’t process language.

With the emergence of deep learning, computers have begun to work on language, translating from one language to another and, in some cases, understanding language too. It brings us to a new kind of AI called “compound” neural networks.

These networks consist of multiple layers; the more layers, the better, so hundreds of these neural networks add up to a complete artificial brain. Once the network train on a data set containing many different data types, it can effectively carry out complex tasks. In the future, we could use them to interpret and process conversations.

Artificial neural networks

When we talk about AI, we usually mean computers that have some AI technology integrated into them.

For example, if we are talking about AI in a smartphone, this could mean:

a) A smartphone with an embedded AI that helps recognize what we see in the world and what we want to see.

b) A smartphone with an AI that will allow the AI technology to recognize faces, voices, and other objects and then communicate with us accordingly.

c) A smartphone with an embedded AI that uses the information to change how the phone will function. For example, it could recognize where we are and then take a step to help us reach our destination quicker.

The developments that have taken place recently in AI refer to as “deep learning” technology. The general idea behind the technology is that a computer can learn complex tasks by analyzing large sets of data. The more data you feed into the computer, the more intelligent it will become. In the same way that we can teach our computers to recognize objects, we can also teach them to identify the languages we speak.

For example, as we have recently written about, companies like Google and Amazon are now making computers that can decipher spoken English.

The previous way of teaching computers to understand languages was called “conventional computer language” or “text-based AI .”Conventional computer language is, of course, the kind of AI we use every day, allowing our phones to recognize the language we speak, for example.

The best way to give conventional computer language a rough idea of what we are saying is to write it out. You can use the keyboard or use voice commands. ( Pindawala )

Google Translate

Google Translate is one of the best examples of conventional computer language (text-based AI). The system first needs to be trained. Have we written more about this process elsewhere on Which? Conversation.

You train a conventional computer language model by giving it data. For example, you could use an online speech recognition system to record yourself speaking a foreign language and then use Google Translate to translate what you say into the language you’re talking, usually spoken Russian.

As you can see from the table below, the system train is slightly different from conventional computer language. In the method used to train Google Translate, you feed it audio files of speech, and the system can then “learn” the sound patterns of the language.

Alternatively, you could feed Google Translate a list of words in another language, the same way we would learn a foreign language. However, the advantage of conventional computer language is that you can automatically translate your computer into a foreign language. It means you can avoid making significant compromises when traveling abroad.

IBM Watson

IBM recently made several big announcements demonstrating the progress made in “compound” neural networks. For example, in March, IBM’s supercomputer, Watson, won the US quiz show Jeopardy, beating humans in the trivia game.

IBM’s new generation of supercomputers uses this technology to process information more quickly and accurately. Computers “learn” to answer questions by analyzing large amounts of data. In Watson’s case, this information comes from billions of questions responded to online.

What has been the impact of deep learning in the UK?

This spring, for example, the startup Reality Lab ran a series of consumer trials to teach a virtual reality film-making system to identify objects accurately. So far, the system can recognize objects more than 95% of the time. The next step is to explain its results in a human-readable way.

Deep Mind has also just launched a machine-learning research facility in Edinburgh alongside the University of Edinburgh. It will bring together researchers from the company and other universities, including Glasgow University.

!!!

Thanks for reading. Visit the Site for more articles.