With the growing emphasis on artificial intelligence (AI) VDC News thought the following explanation of what AI is and its different forms might be useful to readers.This article from tech company, IBM, provides a useful typology for understanding different types of AI by looking at its capabilities and its functionalities.
Let’s start firstly with the three types of AI by capability:
- Narrow AI (sometimes called ‘Weak AI’) – this is the only type of AI that is currently available today. It can be trained to “perform a single or narrow task, often far faster and better than a human mind can”.
- General AI (or ‘Strong AI’) is not yet real but scientists theorise that they will be able to build Artificial General Intelligence (AGI) that will be able to “use previous learnings and skills to accomplish new tasks in a different context without the need for human beings to train the underlying models.” In other words AGI will theoretically be able to learn and perform any “intellectual tasks” that a human can.
- Super AI (also referred to as ‘Artificial Super Intelligence’ is also only a theoretical proposition at the moment – but if it can be built Super AI will be able to “think, reason, learn, make judgements and possess cognitive abilities that surpass those of human beings”.
So right now – even though some of what we see being done with AI looks like wizardry – the tools we are using and the output they produce is ‘Narrow AI” and the humans are still in charge.
Turning to AI functionality and again some functions are still theoretical. Today two of the four types of AI functionality are currently in use and two are still just theoretical functions that AI might have in the future.
- Reactive Machine AI – these AI systems do not have a memory and do very specific tasks – they work only on currently available data and can analyse huge amounts of it to “produce seemingly intelligent output”. Examples include IBM’s Deep Blue chess computer which beat the chess grandmaster Garry Kasparov (by analysing the chess pieces on the board and predicting the probable outcomes of each different chess move) and Netflix’s recommendation engine – helping serve you up viewing recommendations based on what you’ve watched before. (Reactive Machine Learning is the type of AI used in personalised learning – which is explained further below.)
- Limited Memory AI – these systems can recall past data and outcomes and monitor “specific objects and situations over time”. They include generative AI tools like ChatGPT, as well as virtual assistants eg Alexa and Siri, as well as the technology in self-driving cars.
- Theory of Mind AI – this is just a theoretical type of AI for now but if it can be developed it would be able to understand the thoughts and emotions of others ie it could “simulate human-like relationships.” It would also be able “to understand and contextualize artwork and essays, which today’s generative AI tools are unable to do.”
- Self-Aware AI – VDC News’ ghostwriter hopes she won’t be alive if Self-Aware AI is ever introduced, it sounds scary. If it can be developed (and scientists are currently working on it and the other theoretical types of AI) it would be able to understand “its own internal conditions and traits alongside human emotions and thoughts. It would also have its own set of emotions, needs and beliefs.”
For now as we grapple with the forms of AI that are currently available – we need to look beyond generative AI and look at how other forms of AI can and are playing a role in education, like personalised learning.
Personalised learning is defined as “an educational approach that customizes learning experiences to address each student’s strengths, needs, skills, background, and interests.” In education systems which include online learning, many educational developers have used machine learning to help each student have a personalised learning experience (situated with their course curriculum). While personalised learning has been more readily deployed in schools and higher education – with their greater uptake of online learning – in the UK and Europe it is being introduced into vocational education systems.
A good example of personalised learning’s demonstrated ability to lift student attainment is in Singapore’s school system. The government of Singapore trialled personalised learning in a pilot which was subsequently evaluated by one of Singapore’s universities. Because personalised learning can improve student learning (students remember more of what they have been taught and retain that knowledge for longer), based on the results of the pilot personalised learning has now been rolled out across the Singapore school system, as part of the country’s AI strategy.
