The Easy Read Toolbox dreams of a world where it is normal, and indeed expected, to provide information in accessible formats. Our members can access a wide range of support to upskill them in Easy Read and Plain English. Our team of writers and illustrators translate complex documents into different formats. We work with clients to tailor our support to best suit their time, budget, and individual contexts. Even then, we know some businesses that need extra support but cannot afford it. 

What can we do? We’re committed to paying our team fair working rates. So, we need innovative ways to work smarter and faster. We’re exploring the possibilities of generative AI – which is explained in this link. This blog post outlines our current approach and ethics. 

Transparent use

We will always be open about when and how we use generative AI. We’re currently exploring two main uses:

·      Self-management: We predominantly hire and contract people with disability. We’re supporting them to discover how modern technology can help support them, for example by training virtual assistants to offset the impact of memory lapses and distractibility. 

·      Cheaper options for clients: We’re training a virtual assistant to create simplified drafts of complex documents, which our team then checks for errors, inconsistencies, and hallucinations. The produced document may not meet accessibility requirements but represents a step forward in inclusivity. It will be added to our list of potential services.

Privacy

We respect clients’ privacy and confidentiality in all aspects of our work. With regards to the use of generative AI:

·      Clients will not have their documents, or sections of them, provided to the artificial intelligence programs without their knowledge or consent. It is not a necessary part of contracting our services. 

·      When clients contract for the AI service, we will ensure that personal and sensitive data is not uploaded to the programs. Although many programs do not store data verbatim, we still recognise the potential risk.

Complementing not replacing

Creating fully accessible documents, such as Easy Read, is still fundamentally a human endeavour. It often requires creative reworking and problem-solving informed by a deep understanding of the specific audience's needs. Our team will still have many opportunities to grow and develop their craft. 

·      When our writers review first drafts produced by artificial intelligence, they use higher-order thinking. Our pay rates will reflect the increased cognitive load.

·      We are committed to not replacing the work of artists and photographers. We will not add images created through artificial intelligence programs to our image bank.

Ethics and representation

Artificial intelligence is likely here to say. By using it proactively as a business, we are joining the conversation on how and when it is appropriate to use it. We want to model how to incorporate these technologies without asking people to simply be more productive at the same pay rates.

We are also helping to increase the diversity of experiences these programs are training on. They are only as good as the input they get, and the current programs have inherited biases from their source material. That is another reason why skilled human oversight is so crucial.

Concluding remark

When I asked ChatGPT about the ethics of using it in an inclusive workplace, it produced a paragraph which will leave as a conversation piece at the end of this blog post, as well as this link to UNESCO’s video on responsible governance of AI. 

“The ethical use of ChatGPT in the workplace is not just about integrating a new technology; it's about shaping a future where AI enhances human capabilities, promotes inclusivity, and opens new avenues for creativity and employment.”

OpenAI. (2024). ChatGPT (Feb 01 version)

 Written by Kathryn Carter, for the Easy Read Toolbox.

Simplified PDF Version - download here