1. AI agents come into their own
Unsurprisingly, we’re likely to see further important developments in AI. One area that’s likely to take off is the use of AI agents. These are intelligent programs that are given objectives by humans and work out the best ways to achieve those objectives. Agents can write computer code, which could have a big impact on the way that tech companies work and could allow people without advanced coding skills to develop programs, apps or games.
2. Customisation, with help from AI
In education, the focus has traditionally been on linear programmes of study, with pre-determined entry and exit points lasting a number of years. Imagine a course of study that is uniquely tailored to individual students based on their experience, skills and abilities. Bespoke degree programmes centred around the student are already being explored in the US with AI.
These are not just bespoke with regard to the content and curriculum, but also in recognising the special needs of the student or indeed how the learner may feel at any one time. This can include AI that adjusts the learning activity and study based on how much sleep you had last night, which is linked to smartwatch data.
Education isn’t the only area where AI could help with customisation. The management consultancy Accenture suggests that private companies will be able to train their own, custom large language models, the technology behind AI chatbots such as ChatGPT. These could be trained with data specific to particular business areas, making them more effective for those firms. But these companies would have to use billions of pieces of data. We’ll see progress towards this objective in 2025.
Small language models (SLMs) are being developed to perform precise tasks more efficiently. They don’t need to be trained on as much data and require less computing power. This means they can be used more easily on so-called “edge devices” – smartphones, tablets and laptops – without relying on computing resources hosted in the cloud.
Towards practical quantum computers
Developments in quantum computing could lead to machines that can solve complex tasks that are beyond the capability of most classical computers. Researchers have moved away from trying to break records for the number of basic processing units, called qubits, and towards correcting the errors that quantum computers are currently prone to. This is a step towards practical quantum computers that have some useful advantage over classical machines.
iPhone 16
Small language models allow AI to be more easily used on edge devices like smartphones. Raman Shaunia / Shutterstock
You must be logged in to post a comment.