I recently wrote an article for the ANME on ChatGPT and on the benefits but also risks.   You can read this here.   My view is that AI models like ChatGPT are going to become all the more common and also more and more accurate, and therefore we need to explore them and identify how they might be positively used within education.   Seeking to block their use is, in my opinion, guaranteed to fail.   

Following my post, I saw a reply on twitter to the article with ChatGPTs view on AI and education.  You can see this here.    It picked up a couple of points which I hadnt included in my piece and I note that some of my piece actually included content generated by ChatGPT itself.    It wasn’t obvious that ChatGPT had a hand in both pieces which suggests it wont be easy to identify where ChatGPT is used.

All this got me thinking about how ChatGPT might benefit IT Services and the IT teams particularly in schools.   As such I gave some quick thoughts as to possible uses cases, which I have outlined below:

User guides and Help

ChatGPT can be used to create a knowledge base of information that can be easily accessed by IT staff and other school personnel including simple user and help guides.  This seems like the most obvious and easiest use of ChatGPT;  I have already tried asking it some questions in relation to iPad related issues and its responses were clear and accurate.

Creating software and other solutions

Where schools are creating their own internal software solutions including website solutions, ChatGPT can help with the basic code building blocks, thereby speeding up development.   It will still require human input to finalise the projects and add that bit of creativity and flair however ChatGPT can get us part of the way there, thereby saving time and resources.

Policies, processes and procedure documentation

Writing policy and process documentation can quite often be a long and laborious job but ChatGPT and other AI language models can quickly put together a basic document which human staff can then refine and customise to fit the school.

Chatbots

ChatGPT can be used to create a chatbot that can interact with students and staff, answering questions and providing information.   This therefore allows IT support staff to focus on more complex issues or more strategic tasks.

Language Translation

Where schools include non-English speaking students ChatGPT can be used to assist IT support staff in communicating with non-English speaking students and families by providing translations in real-time.

Process automation

A number of the above relate to process automation where ChatGPT is used to automate common support tasks, such as answering frequently asked questions, troubleshooting basic technical issues, and providing instructions for software and hardware.   There are likely other areas where simple processes can be automated through the ChatGPT or other AI Language models.

Conclusion

I think one of the key conclusions I arrive at from my thinking is not related to the benefit of using ChatGPT, or other AI language models, in itself, but for the potential for ChatGPT and a human user to work together.   This hybrid approach of AI and human is, in my view, the way forward as both complement each other.  The AI solution can easily do the basic and repeatable parts of a task, such as creating a user guide, while the human can bring that flair and creativity to make such guides engaging, accessible and usable.    It isnt a case of ChatGPT or humans, or ChatGPT replacing humans.

I suspect there are many other applications of ChatGPT within an IT Support or IT Services capacity which are yet to be realised and I look forward to finding out more in terms of how AI Language Models can enable IT staff to deliver, enhance and even redefine the services provided to users in schools and colleges, and to the communities they serve.

These are interesting times!

Meta released a chatbot for use in the US where its responses are based on internet based data.   It wasn’t long before the chatbot was being less than positive about Meta’s CEO Mark Zuckerberg.   Overall, a bit of a novelty but it might also give us a little bit of insight into the Artificial Intelligence or Machine Learning algorithms which underpin an increasing number of the services we use online.

It is highly unlikely that Meta specifically programmed their chat bot to suggest that the CEO did “a terrible job in testifying before congress” however this is the feedback it provided upon being asked “what did you think of mark Zuckerberg”.    This response is likely the result of the chatbot analysing data sources on the internet and identifying this response as most likely to be true, or at least true in the perceptions of those sharing their thoughts online.   So here we see a couple of problems:

1. As users and even developers, we will not necessarily be able to identify how the response was arrived at.   It’s a black box system;  We can see the inputs and the outputs but not the process.    Considering this should make us a little bit nervous as, especially for important decisions, it would be nice to understand how the answer an algorithm provides was arrived at.   Imagine an AI being used in assessing mortgage applications;  How would you feel if no-one can example why your application was refused?    From a user point of view, as a black box system, there is also the danger that the service provider does have control over the algorithm and therefore can directly influence and control feedback to suit their own needs.  In this case the black box system provides a smoke screen for potentially unethical practices.
2. The chatbot repeats what it sees to be true or the commonly held belief, based on the data sources it accesses.  Bias could easily be introduced here through the internet sources which the chatbot is provided access to or through the queries it might use in identifying pertinent information.   We should be naturally questioning of a solution which may be inherently biased.   One example of this is the issues surrounding facial recognition where the AI was trained largely on white rather than coloured faces, due to the predominant skin colour among those developing the AI solution.  As such we ended up with AIs which did a poorer job of facial recognition when presented with faces with non-white skin colour.
3. Again, relating to the repetition of commonly held belief, the chatbot may simply act as an echo chamber for commonly held beliefs, disregarding minority views.    And if a number of chatbots were to be used together they might be able to powerfully shape the truth on social media channels through repeatedly posting.

Some of the above is of concern but then I start to think about the alternative and a human rather than AI based system.    Humans are not transparent in their thinking processes although they might seek to explain how they arrived at a solution, we rely on sub-concious influences and decision making processes to which we have no access.    Humans equally like an AI based system may be biased or may seek to service their own needs or the needs of their employer.    And humans also tend towards the likeminded, which therefore creates the echo chambers mentioned above.    So maybe AI is no more problematic than a human based solution.   

Is the challenge therefore that AI is technology rather than a human being like us?   Is it maybe that this difference may influence our feeling of unease or unhappiness with the risks mentioned above, and that we simply accept similar issues in human based processed because, after all, we are “only human”?

Firstly, AI can cover a very broad range of activities.   Is it AI designed to interpret natural language such as your Alexa can identify and then respond to you verbal queries, or are we talking about a more general AI solution more akin to Commander Data in Star Trek?    There is quite a gulf between these two extremes, with the 2^nd of them likely to be some time off before it is achievable.

If we therefore accept we are looking at using specific focussed AI solutions in schools by 2025 I think they have clearly got the year wrong as we are already doing it now, in 2022.    We have our spell checker and grammar checker in Word, we also now have our transcription tools in Teams and PowerPoint including the ability to offer real time, or near real time, translation of spoken content.  These are all AI or maybe machine learning based solutions being used in schools and colleges, being used by teachers today.   Not 3 years away in 2025, but today.

So, the headline seems on initial inspection to be quite aspirational and inspirational, for teachers to be using artificial intelligence in their classrooms in only 3 years time.   But a more detailed look and we find it isnt so inspirational as we are pretty much already there.   Maybe the headline hints to a greater use of AI or more advanced AIs being used more often and to greater effect but that’s not the way the headline comes across.   Maybe we will use more AI based platforms, such as learning platforms which direct students through personalised learning programmes, although I have some concerns about this too.  Or maybe there will be greater use of AI and machine learning in the setting and marking of both summative and formative assessments.

I suspect AI use in schools will grow between now and 2025.    I suspect it will grow to be more common in general so wont be a school centric thing, however I suspect that a teacher will still be a teacher and the key to teaching and learning, and the use of AI tools, like the current EdTech tools, will be skilled teachers to wield them as and when appropriate in crafting the best possible learning experience for their students.

Firstly before I get into my views on AI let me first take some issues with the reporting and with the parental perception part of the research.   Looking at the research which you can find here, the question asked of parents focused on the “consequences of the use of AI”.   This feels a little negatively biased to start with.    Under this banner question a serious of sub-questions were asked with the participants asked to respond with either don’t know, fairly concerned/very concerned or not very concerned/not at all concerned.  Again the options hint towards negativity and therefore introduce bias.   And finally the sub question itself in relation to transparency for example focused on concerns relating to a “lack of transparency”, again a negative implication and further negative bias.     It is also worth noting that the survey only had 1225 parents contributing.    I think this falls very short of a sufficient sample to draw any meaningful and generalisable findings.   Despite all of the above the TES decided to pick up and report the findings of “parents’ fear about artificial intelligence in schools” including indicating an “overwhelming majority of parents are concerned”.   I find it somewhat funny that concern of potential bias in relation to AI was reported in an article itself so loaded with its own bias.

So to my views;  I myself have concerns regarding AI use in schools however also see much potential.   Funnily enough the Nesta report to which the TES referred concludes that AI in education “promises much to be excited about.”

Given the negative bias in the TES report lets therefore start with my positive views as to the potential for AI in education.   AI is very good at identifying patterns and divergence from patterns within large data sets.    This makes them ideal for analysing the wealth of school and wider educational data which exists to help educators, those responsible for educational policy and decision making, school leaders and even the teachers themselves.    Now thoughts may instantly jump to achievement data sets resulting from testing, final exams or teacher awarded grading however the opportunities far exceed this area.   Take for example data taken from school Wi-Fi, where students are allowed access, in relation to student movements around the school.   This data might help a school reorganise the school day or restructure the timetable in order to become more efficient and maximise the learning time available.   It might also be used to redesign learning spaces or develop spaces for students to rest, take a break and address their wellbeing.  This is but one example of how AI might be used along with school data.

AI can help direct students to appropriate learning materials using data to identify the areas where students need additional support along with the best support materials to meet these needs.    Some platforms already exist and are exploring this opportunity including Century, a platform which I heard very positive stories regarding when recently speaking to students at a school using it.   Platforms like this might prove highly valuable additional resources to complement classroom teaching or to provide a more effective homework platform.   This area and use of AI is likely to continue grow with the development of more and more online learning content being key to this.

AI can help with teacher administrative tasks such as registration conducted via facial recognition or marking of tests by natural language AIs that can apply a given marking criteria to student submitted work.    We also need to recognise some of the AIs that are already available including voice recognition and dictation, which is now a feature of the MS office products.    Googles search facilities, a now standard feature used in schools and classrooms, also quietly uses AI yet we don’t bat much of an eyelid to it.

The negatives implications which exist in relation to AI generally apply beyond the educational context, albeit the educational context in teaching our future generations makes things all the more worrying.

AIs need to be taught and to learn with this done using training data sets.   The worry is that bias in the training data set will result in bias in the AIs decision making.    As a result an AI which was developed in the UK, and therefore trained using UK based data, and used successfully in UK schools may not be appropriate for use in schools in Asia or the Middle East due to its decision making being biased towards a UK context.   That said, this same issue would impact on any product or service, or even individuals where they seek to operate outside their normal context.   We all have an inherent bias, we “humans”, create the AIs and train the AIs so is it realistic to expect an AI without bias?  I suspect part of the issue is a concern in relation to a particular bias being introduced purposefully however I think it is more likely bias in AIs will arise accidentally as it general does within humans.

There is a concern that AI decision making based on large data sets may become impossible for humans to explain or understand, as the decision making process could be based on huge amounts of data.     This brings with it the concern that we may lose some of our control.   If a teacher recommends a career track for a student they will be able to explain how they arrived at this however if an AI was used, the teacher may be able to present the AIs findings but may be unable to explain or understand how this was arrived at.   How many parents who be happy with a suggested career path for their child without any explanation available?

Linked to the above is a concern of “determinism” where AI might identify an end point and then through its actions lead to this occurring.  So those students identified as achieving a C grade in GCSE might be presented with content and learning materials which lead them to achieve exactly that.  This concern is again about a lack of control however it could be suggested we are deterministic in some of the practices already in use widely in schools.   Take for example the setting of students into ability bands, is this not potentially deterministic as the students in the top band get the most challenging content which may enable them to achieve top grades while the students in the lowest band gets easier materials which means the don’t learn the more complex materials, and as a result are unable to achieve the top grades.    Also is there a danger of determinism every time a teacher reports a predicted grade to parents or where a school uses ALIS or other benchmarking data?

Overall AI is going to find increasing uses in schools.   My gut feeling however is that for the foreseeable future this will be very much in a subtle way as data analysis systems start to suggest areas to investigate within school data, accessibility tools including dictation and translation support students in class and AI driven learning platforms provide personalised learning opportunities beyond the classroom.   These are but a few examples of things already happening now.  These uses of AI are likely to become more common.   Discussion of AI reminds of a quote in relation to effective technology integration being such that the teacher and learners don’t even stop to think about the fact they are using tech, the tech use is transparent.   I think AI use is going to be exactly this, and the AI in Googles search goes some way to provide this;  When was the last time when you were conducting an online search that you stopped to think about how google search works and how AI may be involved?