This section focuses on individual lecturers and teachers, particularly those in assessment roles. Its focus is on a series of recommendations framed as 'Dos' and 'Dont's'.

The overarching aim is to promote AI literacy as well as responsible and beneficial use of AI in education. It is important to note that AI literacy is something that we will all need to grow and that even those with extensive computer science expertise are still finding their way through the implications of generative AI on teaching and assessment practices.

If you are a module leader, the chances are you will have (or have planned) discussions with programme leaders and colleagues who teach on the same modules as you. If not, please do ask when these conversations will occur. Programme and module leaders will be thinking about the broad vulnerabilities of assessments, and it will be useful to think carefully about how we progress. It is important however that our messaging is clear and consistent so please ensure you have familiarised yourself with the university-level (macro) and programme-level (meso) guidance.

At King’s we are committed, along with all our Russell Group partners, to follow shared principles:

• Universities will support students and staff to become AI-literate
• Staff should be equipped to support students to use generative AI tools effectively and appropriately in their learning experience
• Universities will adapt teaching and assessment to incorporate the ethical use of generative AI and support equal access
• Universities will ensure academic rigour and integrity is upheld
• Universities will work collaboratively to share best practice as the technology and its application in education evolves.

The key implication is that generative AI technologies and functionality are developing swiftly and, by definition, are the worst they will ever be right now. The following approach and recommendations may be of use, especially to those who have only had limited exposure or experience of using generative AI. The guidance and heuristic have been developed in partnership with KCLSU.

We do of course acknowledge the workload implications whenever there are changes or innovations like this, especially at the start of a new academic year. However, we hope that irrespective of starting point and across all disciplines, colleagues will be able to find time to engage with generative AI at least minimally.

This iterative model’s launch position is the AI landscape in which this guidance and the King’s FutureLearn course should be seen as essential starting points. The 'Dos' and 'Don’ts' below may suggest an appropriate starting point for colleagues who have yet to explore much in the realm of generative AI. Needs will vary greatly according to discipline, but some degree of experimentation and a review of assessments would also be strongly recommended. The ‘Do’ recommendations that follow are clustered according to the stages of the model:

Familiarise/research the AI landscape

• Reflect on the extent to which AI (perhaps less visibly) already permeates our work (e.g. automated meeting captions, transcription, grammar and spell-checking tools, automated alt-text and other accessibility functions)
• Consider the accessibility implications raised by use of generative AI tools: how could they create or remove barriers?

Experiment with tools

• If you have yet to try out generative AI tools, experiment with free versions (e.g. Google Gemini, ChatGPT, Claude AI) and consider the sheer volume of new Generative AI tools available. Please do try Microsoft Copilot (you need to be logged into your KCL account for protected search and increased data security).
• In text-based generative AI tools, experiment with productivity utilisations such as summarisation. (Note: whilst these tools may offer a ‘summary’ based on supplied web links these are notoriously unreliable so for summarisation, experiment with copying text into the tool or using a bespoke tool such as Humata or explore how Claude AI allows easy document upload).
• Share successes in novel or existing assessment designs that could be transferable (within teams, in faculty and across the university via King’s Academy).
• Consider the importance of prompt engineering and support your students at appropriate levels to consider the following:
  • Clarify in your own mind the desired output as precisely as possible
  • Define your audience
  • Clarify tone, style, length and format
  • Adjust or improve for specificity

Consider impact on course content and progression routes

• Discuss with colleagues how AI is impacting likely progression routes (employment or further study) and the implications for curriculum content
• Be open to potential impacts on students or colleagues in all their neurodivergence and staff and students with disabilities, including those who may already have begun to depend on generative AI to support their workflows
• Consider how the PAIR (problem, AI, interaction, reflection) framework might support integration of AI into your curricula

It is also worth considering ways that generative AI might improve your in-course productivity and/or resources, such as: 

• Creating copyright-free novel images tailored precisely to need
• Turning rough notes into full prose - e.g. for extensive feedback
• Translating text for research or to communicate with international partners
• Idea generation for presentation, research topics, vignettes or case studies
• Quiz creation or question bank generation
• Creating jargon lists or glossaries
• Basic assessment rubric creation
• Generating boilerplate feedback based on jotted notes and rubrics
• More tips from King's Academy

Review assessments

• See how Generative AI tools fare with your assignments or similar tasks
• Consider the extent to which your assessments are knowledge-based and/or are dependent on factual recall
• Review your handbooks and formal documentation to assess the extent to which you have built in flexibility in assessment tasks and overall assessment design
• Ensure you have a clear statement about acceptable use and have shared the acknowledgement statements
• Discuss within teams and with students how module and programme level outcomes could be assessed authentically, fairly and with integrity and how this aligns with existing/proposed assessments
• Consider ways in which to de-emphasise assessment products (e.g. the essay or lab report) and emphasise focus on the process of producing the assessment.

Communicate/work with students

• At induction or in new year group tutorials, discuss generative AI in particular and acknowledge current anxieties - engage students in hype vs reality discussion if appropriate
• Talk openly with your students or supervisees about AI broadly, especially in relation to innovations in associated career paths and on ethical and critical use. Discuss and share ideas and approaches to practical uses of Generative AI
• Acknowledge wider ethical issues and the legitimacy of colleague and student objections to using nascent, not thoroughly tested, data security-uncertain technologies

• Panic! It’s important to acknowledge the likelihood of perhaps profound changes but these challenges are not unique to us, and the more we can collaborate across the sector and with our own students the more fruitful the journey is likely to be.
• Assume that all assessments will need a radical overhaul 
• Assume that students will be conversant with generative AI tools and the implications of their use
• Forget to acknowledge your own use of generative AI
• Pay for any tools unless you have explored alternatives and are fully aware of cost and data implications
• Ask students to purchase subscriptions for any tools
• Copy and paste student work into tools like ChatGPT (it will NOT reliably report whether it produced that text)
• Overestimate Generative AI capabilities or anthropomorphise these technologies: they should complement what we do not replace it
• Use unsupported web-based tools that purport to detect AI generated text (unless for experimentation with text that is yours or freely available online)
• Assume any text generated will be factually accurate
• Ignore potential for biases in outputs

Whilst our focus currently - and understandably - is on the ways we might modify and improve our assessment practices, it is also worth considering how in both the short and longer terms, generative AI may change how we work and the ways in which we can optimise productivity.

A key aspect to our ongoing and rapidly growing understanding of both the workings of generative AI and the implications of it is that we are all learning, so it is vital that we communicate with one another. A golden rule for all staff is that we should centre dialogue amongst ourselves and with our students. If you are uncertain and the answer is not immediately apparent in the resources here, then speak with your colleagues and encourage students to do the same.