The world is on the brink of a revolution in education, as the likely impact of artificial intelligence (AI) in schools is becoming increasingly clear. While some AI-powered tools are already in use in a small percentage of classrooms, we have not yet seen anywhere near the full potential of AI in the sector. Indeed, the exponential pace at which AI technology is suddenly advancing suggests that very soon AI will become an essential part of the learning experience for students all over the world. Before long it will likely be the dominant feature in teaching.
This article briefly explores the importance AI will have in helping pupils develop their skills, the benefits it will bring to those who have been neglected by the current system, and explains the tremendous pace at which all of this will arrive. As the world embraces the bold and adventurous new world of AI, it is important that educators and policymakers stay up-to-date with the latest developments to ensure that their students can benefit from this technology to the fullest extent.
[1] The Rapid Pace of AI Evolution
Even most experts seem to have been surprised by the rapid pace of development in AI over the last few months of 2022 and into the first quarter of 2023. This seems to have been driven both by some amazing technical advances and some market dynamics. In short, competition between large corporations and the addition of ‘open source’ players into the market has meant that some previous caution has been abandoned as entities compete for market share.
There are many developing areas which are directly relevant to mainstream education, and no doubt educational establishments will have already noted that there should be huge savings to be made in administrative areas, and teachers will be hoping that AI will help them with the huge paperwork requirements of lesson planning and assessment of student work. All of that is true, and it should arrive once AI can be integrated into existing systems. That is not where AI will have the most immediate impact however, which will instead be when AI and students interact directly as below.
[1.1] Text to Image AI. This is here now. Already at a low cost (or even zero cost if run locally). Tools like Midjourney, Stable Diffusion or Dalle-E run in a web browser and can create high quality images from a simple text prompt. While initial versions were of no threat to professional artists, the latest versions are already immensely impressive. Certainly most students can suddenly produce far more powerful images than they would ever have been capable of previously.
Think of this as AI’s ‘gateway drug’ for many students and their parents. It is the application which is immediately understandable and usable.
[1.2] Chat and Search AI. Most students are given homework, and often this involves writing a report of some kind. Tools such as ChatGPT are able to produce such things in a few seconds, and rewrite them as often as needed to cover teacher requirements. They will also soon be available at absolute minimum cost, many running entirely on a phone or laptop.
In the short term this is the arena for a battle between schools and students as teachers worry how to assess an essay that may have been entirely written by an AI. Of course, this has always been a problem where students have a helpful parent or older sibling. It will soon become obvious however that schools must teach students how to work with AI, in the same way that they have already been taught how to work with books or equivalent references.
Note that both of the above types of tool are available today and that there are more appearing each week, either as original software or as variants of the above. Nobody should ever be thinking of signing up for an annual fee for these tools — there will be another competitor along before long offering more capability for less money.
Section 2. The particular relevance to the neglected student
Our system is often said to work reasonably well for the student of average ability but to fail both the most talented and the least able. To balance this any well funded education establishment provides extra support in the way of ‘classroom assistants’. The children of better off people may also have personal tuition, or they pay for schools with smaller class sizes.
Suddenly there will be a new option, and it will probably first grow at scale in the retail market. As soon as good AI Personal Tutors are offered they will spread at an amazing rate. After all, they will be available whenever required, at a cost which is a fraction of that of a real tutor, and they will be able to tailor their teaching to the individual student and their interests.
As a simple example, here is today’s ChatGPT responding to a request I made.
Imagine now how a slightly more advanced version of ChatGPT with specific knowledge of the required curriculum and of the student’s existing knowledge and abilities would function. Add on the capability to be able to understand and respond to spoken questions, and to invisibly make any necessary assessments of student ability and participation levels.
AI will also be able to support the development of soft skills, such as teamwork, communication, and problem-solving. AI-powered platforms can easily create collaborative learning environments, where students can work together on projects and receive feedback from their peers and teachers in real-time.
Section 3. Why schools should engage quickly
Within a few months, all of this will available in the student home. If a school decides that it will not use such systems it risks being marginalised, not least in the eyes of its students. Why would a student pay attention to a teacher who delivers content using only traditional tools. Certainly parents will soon work out that the route to examination success and college access lies through full use of AI tutoring. While they would no doubt prefer that to be offered as part of the normal school education, they will make their own arrangements if necessary.
The biggest danger for the authority of traditional schools lies in overcaution. That means they need to work out how to implement AI tools despite AI’s known flaws and weaknesses, for example:
[3.1] Hallucination. AI systems write great fiction, and they can also tend to invent facts when they cannot locate the accurate answer in their usual sources.
[3.2] Bias. All AIs base their analysis on the documents they have absorbed, and the training they have been given by their creators. This leads to inbalances in their expression of political views and in how they understand many social issues.
[3.3] Unpredictability. An AI is not a pre-programmed tool that always produces a set answer when responding. By its nature there is a level of randomness in the precise answer it gives to any general question.
[3.4] Data protection. In the fast moving world of AI development, not enough attention may be given by all providers to issues around ownership and protection of student data.
Add to this the question of students passing off AI produced work as their own and schools will naturally wish to proceed carefully. This should not stop them progressing at a good pace however. The benefits to be gained here are substantial, so it is essential that all involved work together to quickly provide the best possible learning experiences for all students.
To conclude, AI has arrived with enormous potential to help pupils develop their skills, whatever they may be. By providing personalized support and feedback and creating collaborative learning environments, AI can enhance the learning experience for students and prepare them for success in the future.