Key Points:
In this post, I share my journey of discovering how Artificial Intelligence (AI) is revolutionising Speech and Language Therapy (SLT). I explore:
The SLT crisis: I highlight the current challenges, including long waiting lists and staff shortages.
My AI journey: I recount my 18-month exploration of AI tools, revealing their potential to dramatically improve efficiency and effectiveness in SLT practice.
AI in action: I describe ten specific roles where I've found AI to be an invaluable 'non-human co-worker'.
Ethical considerations: I address some of the important limitations and ethical considerations we must navigate when using AI in healthcare settings.
Future possibilities: I share my predictions for AI's future in SLT, including its potential impact on waiting times and therapist workload.
Getting started: I offer practical tips for fellow SLTs interested in incorporating AI into their practice.
Through this piece, I aim to provide both inspiration and practical guidance for SLTs looking to harness AI's potential to enhance patient care and professional efficiency.
The SLT Crisis: Numbers that Demand Innovation
Spending the first 22 years of my career within the NHS gave me a front-row seat to the challenges of long waiting lists and staffing shortages. In the seven years since transitioning to independent practice, I have also witnessed the impact of COVID-19 and the continued lack of investment in SLT, which remain persistent barriers to providing timely and effective care.
According to the NHS Community Health Services Waiting List Data for May 2024, there were 76,331 children waiting to be seen for SLT, with 6,053 of them waiting for over a year. While some families seek independent SLT, many lack the financial means, and those who can afford it often struggle to find available therapists.
As well as the overwhelming need for services, in April 2023, the Royal College of Speech and Language Therapists (RCSLT) published concerning news of vacancy rates in the profession. Within the NHS and independent sector there was an average vacancy rate of 25% in children’s speech and language therapy services in England, and 24% in adult services. The document states,
“There is no quick fix which can resolve the situation in the short term. However, there are short and medium-term actions presented here that can lessen the impact. We are ready to help the profession innovate and transform.”
There are a range of innovative and comprehensive actions identified in the document but there is no mention of the potential of Artificial Intelligence. My assumption is that this reflects the rapid emergence of generative AI, with ChatGPT only entering the public space a few months before this report was published.
Edit 24/07/24: Coincidentally, on the day I wrote this blog, the RCSLT published an updated vacancy survey highlighting persistently concerning rates of staff shortages. Steve Jamieson, the RCSLT CEO, stated:
“Speech and language therapy services are close to breaking point with high vacancy rates and not enough people in the workforce to fill the roles...The speech and language therapy workforce crisis must be addressed urgently, and we look forward to working with the new Government to find sustainable solutions.”
My intention here is to share my experiences as an early adopter of AI technology and explain how it is transforming my practice. I'll provide practical examples of everyday tasks that have been enhanced by AI in terms of both speed and quality. I predict that AI will significantly improve the efficiency and effectiveness of our work in the future. Whether you're an NHS or independent therapist, I hope this post inspires you to explore AI's potential in your practice.
My AI Awakening: From Novice to Innovator
My journey over the past 18 months has led me from knowing nothing about generative AI, to using it in nearly every aspect of my work. I have developed a passion for exploring its potential and I have familiarised myself with a range of AI tools, all powered by Large Language Models (LLMs), notably:
The speed and creativity of these tools amazes me daily as they create inspiring images and stories, write clear and accessible explanations and answer complex clinical questions with ease.
Large Language Models
LLMs are sophisticated AI systems at the cutting edge of natural language processing. Trained on huge datasets from diverse forms of written language (academic papers, books, online articles and social media posts), LLMs have developed an intricate understanding of human communication. This extensive training enables LLMs to grasp the grammar and semantics of language as well as more nuanced aspects like idiomatic expressions and cultural context.
The beauty of these AI tools lies in their versatility. For SLTs, they offer a range of applications, from streamlining administrative tasks to enhancing therapy session preparation and even assisting in clinical decision-making. However, these tools are not meant to replace the human element in therapy, but rather to augment our skills, allowing us to focus more on our core purpose and contact with service users.
The key to successfully using AI in any profession is knowing how to communicate effectively to achieve the best results. Effective prompting of AIs relies on strong communication skills, making SLTs ideally suited as excellent prompt engineers. I view AIs as my non-human coworkers who have endless patience and time to assist me in my work. I've found that I can complete some tasks ten times faster, and the results are often better. This isn't about delegating care to a robot; it's about collaborating with AI to work more productively.
Your New Digital Assistant: 10 Ways AI Transforms SLT Practice
Here are some examples of how working collaboratively with AI plays out in my day-to-day tasks:
1. AI as a Writer
AI has become my writing partner for blogs, leaflets and training content. The ChatGPT app allows real-time voice collaboration, enabling me to speak my thoughts before committing them to paper. The AI then helps me to organise these thoughts into text.
Gemini excels at generating imaginative stories and factual passages. If I want the story loaded with a specific phoneme, I can just ask.
2. AI as an Editor
When I'm getting stuck explaining a complex concept, I ask AI to rephrase sentences or improve paragraphs. Claude AI, known for its ethical training and nuanced writing, is particularly helpful here. I often cut and paste a piece of writing into Claude and say, ‘improve this’. It can rephrase my words to and help me make more sense in matter of seconds.
3. AI as an Idea Generator
When I’m in need of some fresh ideas, I have asked AI for example to, ‘give me some fun and different methods to target prepositional language with a 6-year-old’ or ‘please review this training course and tell me how I can make it more interactive’ (helpfully you can upload documents to most generative AIs and then discuss the contents).
4. AI as a Research Assistant
AI helps me quickly gather information about clients' specific interests, helping me to connect more easily with a Pokemon fan or a Minecraft enthusiast and making sessions more engaging.
It's also useful for summarising recent publications and articles. ChatGPT has specialised GPTs such as 'Scholar GPT' that can help me to critically evaluate research.
Perplexity is an excellent tool for searching the internet as it can simultaneously search many websites, summarise the content and provide citations for me to review.
5. AI as a Designer
AI image generators are powerful tools for creating engaging visuals for therapy resources and presentations. However, it's crucial to be intentional about representation. When generating images of people, I've learned to specifically prompt the AI to create diverse representations. This includes requesting a range of ages, skin tones, body shapes, and abilities. By doing so, I can ensure my materials reflect and celebrate the diversity of my service users.
I have also used ChatGPT to design an app for teenagers with Developmental Language Disorder to learn vocabulary. It's called 'WordMango' and is free to use. Feedback from young people and SLTs has been positive, with teenagers loving the fact that they are using AI in therapy.
6. AI as an Organiser
AI helps me structure tasks, set to do lists and construct spreadsheets. When I’m faced with a blank page and not sure where to start, I can use it to outline the task and break it down into smaller steps. If I am running a training course I can upload my slides and notes and it will estimate how much time each aspect needs.
7. AI as a Simplifier
AI can be used to summarise and simplify any document, video or website I need to understand. I’ve also used AI to create explanations of complex topics by asking it to consider the intended audience. For example, I asked Claude to explain cochlear implants keeping children, parents, teachers, and SLTs in mind; it gave four very different explanations, appropriate to for each group.
8. AI as an Analyst
I’ve used ChatGPT to analyse anonymised language samples and we’ve discussed appropriate targets and activities. It can also review feedback from trainings, analyse what is working well and suggest possible improvements. However, SLTs are better than these LLMs at analysing speech (a more specialised AI is needed for this kind of collaboration).
9. AI as a Coach
ChatGPT has acted as a business coach, played ‘devil’s advocate’ and conversed in a supportive capacity for me. It can take on the persona I require to guide my thinking. I’ve also sent it a screen shot of an error message on my computer and it has acted as an IT coach, taking me through a process to fix the problem, step by step.
10. AI as a Translator
I can ask AI about the structure of the grammar or sound system in another language. I’ve used it to help make a speech screen in Mandarin (but checked the detail with a Mandarin speaker).
The ChatGPT app can also translate in real time providing an instant way to communicate with clients who speak languages other than English.
In truth, the possibilities are endless, and the limiting factor is more likely the extent of my imagination in thinking of different ways AI could help me.
Ethical AI: Navigating the Digital Minefield in Healthcare
While AI offers many benefits, we need to approach its use thoughtfully and with professional integrity.
Privacy and Client Confidentiality
Protecting client confidentiality is paramount. When using AI, ensure that no identifiable client information is shared. Only anonymised data should be used to prevent privacy breaches. Whilst the AI systems mentioned in this article have strict privacy regulations, this does not mean that interactions will be unseen by human reviewers and engineers or prevent them from being aggregated and used to train future models. Ensure your AI use fits with the policies of your organisation.
Bias
AI models may reflect biases present in their training data, such as gender, racial, or cultural biases. We must be vigilant in identifying and mitigating these biases to prevent perpetuating them in their work. For example, if an AI model is trained predominantly on data from a particular demographic, it may not provide equitable advice for clients from diverse backgrounds. Our awareness and critical evaluation of AI outputs is vital to ensure equitable and unbiased practice.
Hallucinations
AI models can sometimes produce 'hallucinations' – these are responses that appear plausible but are factually incorrect. It is essential to verify AI-generated information before application to practice. Although advanced models reduce this risk, diligence in fact-checking remains crucial. AIs which clearly cite their sources will enable you to fact check more efficiently.
The Human Touch
It is important to recognise that while AI has made great strides, it's not infallible. Tasks requiring complex reasoning, emotional intelligence, or nuanced understanding of context still require human expertise. Although generative AIs seem very human-like in their interactions they not sentient or conscious and they do not think or understand language in the way we do. They are highly sophisticated machines, not beings.
Maintaining Professional Expertise and Accountability
While AI offers significant benefits, it's crucial that we safeguard our core professional skills. Uncritically delegating work to AI risks eroding our expertise. Instead, by maintaining human oversight of all AI outputs, we can expect to enhance our skills rather than diminish them.
This approach aligns with our professional responsibility: we remain 100% accountable for any AI-generated content used in our work. Therefore, it's not just advisable but essential that we critically evaluate and refine AI outputs. In my experience, this process of review and refinement can actually deepen our understanding and sharpen our clinical reasoning.
In essence, the goal is to use AI as a powerful tool that enhances our professional capabilities, rather than as a substitute for our expertise. This approach ensures that we maintain high standards of care while benefiting from technological advancements.
Breaking New Ground: AI's Promising Trajectory in SLT
In speech and language therapy, we can anticipate AI becoming increasingly sophisticated. Advances in natural language processing could result in more accurate speech recognition and synthesis, benefiting clients with speech disorders. The integration of virtual reality may create immersive therapy environments. While these technologies are still on the horizon, we already have access to powerful generative AI and LLMs that we can use today.
Integrating AI into our work could significantly impact waiting times by reducing the administrative workload for SLTs, enabling us to spend more time with service users. Using AI to enhance our clinical work could lead to better and/or quicker outcomes.
I find that collaborating with AI to word targets, strategies, and advice within reports is much less time intensive and often achieves more concise and clear wording. I've also found that incorporating a child's interests into their therapy programme through AI-generated resources has significantly increased engagement. We might also leverage AI to tackle broader challenges, such as redesigning service pathways to incorporate new evidence-based practices.
Throughout my NHS career, I found the most fulfilling moments were when waiting times decreased, service users made significant progress, and I had autonomy in delivering personalised care. These periods saw my stress levels drop, and I felt genuinely empowered to make a difference.
Today, I believe collaboration with AI could be a key factor in improving job satisfaction and retention rates within our profession, bridging the gap while we await the training of more SLTs.
To harness the potential of AI effectively, we need to invest in developing our knowledge and skills with generative AI models. By learning to use it creatively, safely ethically, we can create a more efficient, satisfying work environment that benefits both SLTs and the individuals we serve. This proactive approach to incorporating AI into our practice could be a game-changer in addressing the current challenges facing our profession.
To gain a greater understanding of the current NHS context I attended the Health Foundation's event, "AI in the NHS: Shaping the Future of Health Care," on July 18th. It was reassuring to hear that there is extensive ongoing dialogue about the future role of AI within the NHS. Nell Thornton, Improvement Fellow at the Health Foundation outlined six key strategic priorities for the NHS in harnessing AI in the future (read here). The sixth of these is,
“The healthcare workforce must have the right skills and capabilities to capitalise on AI”.
It has become my mission to become a positive force in achieving this within the SLT profession.
Workshop to Workplace: SLTs' Journey with AI Tools
Since May 2024, I have been running workshops for SLTs to learn about generative AI. I ensure that participants come away with a basic understanding of generative AI, have practical experience of using AI models, explore how AI can be used in their work context and are aware of the limitations and ethical considerations.
My workshops are just a small part of upskilling the workforce, but I am already delighted by the feedback I’m receiving from participants. I’m hearing how SLTs are using AI to write social stories that are perfectly tailored to an individual's circumstances and needs, and in a fraction of the time this would normally take. By sharing the successful AI-generated prompt, teachers are then empowered to create additional social stories on their own.
SLTs who have attended the workshops have expressed how transformative this will be in their work; one SLT recently wrote,
“The best training I have ever been to I think with regards to the potential of helping in my professional life”.
So far, most participants have been from an independent context, but I hope to reach more NHS SLTs as the workshops continue. I'm particularly excited to be presenting at the RCSLT Scotland Connect event on 17th September 2024, with a workshop entitled, “Artificial Intelligence – How can AIs like ChatGPT transform efficiency, staff well-being and client care in SLT?”
Your AI Journey Starts Here: Top Tips for Getting Started
If you're not yet using AI in your practice and are interested in doing so:
Start with free versions of AI tools like ChatGPT, Copilot, Claude, Gemini and Perplexity.
Experiment with using AI for home and work-related tasks, starting with tasks that you know the answers to so that you can evaluate the output.
Chat to the AI like it’s a person, explain clearly what you want and give it feedback. Use my SERVE Prompting Framework to help you craft your prompts.
Always verify AI-generated content before clinical use and be vigilant for bias.
Ensure you avoid any use of client identifiable data.
Get to know the rules and policies for AI use within your own organisation and stay informed of evolving regulations and guidelines.
Continue the Conversation:
I'd love to hear about your experiences using AI in SLT. Connect with me via:
Access to Workshops:
If you would like to join one of my workshops. Visit https://chatterboxsussex.eventcube.io/events for upcoming dates and registration details.
I am also happy to run workshops for groups/teams – please email me to discuss your requirements: rachelbartonslt@gmail.com
Conclusion: Embracing AI for a Brighter Future in SLT
As we stand on the edge of a technological revolution in Speech and Language Therapy, it's clear that AI has the potential to transform our profession in many ways. From streamlining administrative tasks to enhancing our clinical decision-making, AI offers us tools to work more efficiently and effectively than ever before.
However, this potential can only be achieved if we approach AI with a balance of enthusiasm and caution. As SLTs, we must remain at the forefront of this change, guiding the integration of AI into our practice in ways that enhance rather than diminish the human touch that is so crucial to our work.
The challenges facing our profession - long waiting lists, staff shortages, and increasing demand for services are significant. But AI may be the powerful ally we need in addressing these issues.
As we move forward, continuous learning and adaptation will be key. We must stay informed about AI developments, critically evaluate their applications in our field, and always prioritise ethical considerations and client well-being.
The future of SLT is not about choosing between human expertise and artificial intelligence. Instead, it's about harnessing the best of both worlds - combining our unique human skills of empathy, intuition, and relationship-building with the efficiency and analytical power of AI.
So whatever your context, I encourage you to explore the potential of AI in your work. Start small, experiment, and share your experiences with colleagues. The journey of integrating AI into SLT is just beginning, and I, for one, am thrilled to be part of it. Are you ready to join this revolution?
AI Acknowledgment:
In the spirit of the topic of this blog, I utilised AI tools including Open AI’s ChatGPT and DALL-E and Anthropic’s Claude in the creation of this content for idea generation, editing, researching, and image generation. All AI-generated content was thoroughly reviewed, edited, and verified by me to ensure accuracy and alignment with my professional expertise.
References:
NHS England. (2024). Community Health Services Waiting Lists. Retrieved from https://www.england.nhs.uk/statistics/statistical-work-areas/community-health-services-waiting-lists/
Royal College of Speech and Language Therapists. (2023). Workforce Planning in England. Retrieved from Workforce-planning-in-England.pdf
RCSLT. (2024). RCSLT Workforce and Vacancy Survey. Retrieved from https://www.rcslt.org/news/rcslt-workforce-and-vacancy-survey/
Thornton, N., Hardie, T., Horton, T., & Gerhold, M. (2024). Priorities for an AI in Health Care Strategy. Health Foundation. Retrieved from Priorities for an AI in health care strategy.pdf
Comentarios