How have you used AI tools in medical study? by Terese Bird

How I use AI

2025-06-16 13:01:08 UTC

I use AI for multiple of reasons from mainly using it for learning difficult parts of my course where i kind of use it as Personal tutor or to summarise and make me flash cards or to conduct research or find answers in niche topics

2025-06-16 13:01:09 UTC

for quizzes and reinforcing content known, so providing the ilos and prior exam questions for content to generate my own exam questions in a similar style.

ANKI

2025-06-16 13:01:20 UTC

I use ai to summarise textbooks and nice to speed up making my flashcards for ukmla conditions.

2025-06-16 13:01:25 UTC

I have used it to get explanations of different topics which I don’t understand. Summarise notes and make them into flashcards to import into anki

Creation of OSCE Stations/scripts

2025-06-16 13:01:50 UTC

Due to lack of resources, using ai to develop osce stations on specific topics/ station types

Using AI in study

2025-06-16 13:02:29 UTC

Used AI a lot more in preclinical years just because I had more questions than the lecture slides could answer, and sometimes the way a lecturer explained a concept would not make sense to me, so almost had to train ChatGPT to explain medical concepts in a ‘simple to understand-‘dumbed down’ way. In clinical years I’ve used it more for consultation of guidelines and also some OSCE practice. Brief use for revision advice and techniques

2025-06-16 13:03:21 UTC

2025-06-16 13:03:36 UTC

2025-06-16 13:04:09 UTC

pass

2025-06-16 13:04:11 UTC

2025-06-16 13:04:42 UTC

2025-06-16 13:04:43 UTC

AI powered

2025-06-16 14:04:55 UTC

🔹 What is an AI‑Powered Chatbot?
A conversational agent using NLP and machine learning to understand user input and respond naturally—far beyond simple rule-based bots.

🔹 Top AI Chatbots & Their Uses:

ChatGPT (OpenAI): Versatile—writing, coding, creativity, image generation.
Google Gemini: Multimodal assistant integrated with Google Workspace.
Grok (xAI): Edgy, social-media style assistant embedded in X (Twitter).
DeepSeek: China-based open-source chatbot app—cost-effective and quick to train.
Microsoft Copilot: Productivity assistant built into Microsoft Office and Bing.
Character.ai: Custom, personality-driven AI “characters.”
Replika: Virtual friend/companion for emotional support.

Voice based

2025-06-16 14:05:23 UTC

A voice based chatbot is a program that uses speech recognition and speech synthesis to allow users to communicate using spoken language and responds with a human like voice. Eg. OpenAI voice mode

Rule-Based Chatbots

2025-06-16 14:05:49 UTC

Definition:
A rule-based chatbot follows predefined rules and logic (like decision trees or if-then conditions) to answer specific user inputs. It does not use AI or machine learning and cannot adapt to new questions.

Responds based on keywords or button choices
Doesn’t learn or improve over time
Works well for simple, repetitive tasks
Fails with unexpected input or complex queries

Example:

User: "I want to book an appointment."
Bot: "What type of appointment? (1) GP, (2) Dental, (3) Eye Test"
User: "Dental"
Bot: "Please choose a time: 10am, 2pm, or 4pm."

→ If the user says "I need help with my teeth," the bot won’t understand unless that exact phrase is programmed in.

Menu + Button based AI chatbots

2025-06-16 14:07:43 UTC

Essentially, they're relatively simple 'chatbots', defined as 'a type of chatbot that guide users through a conversation using clickable options (menus or buttons) rather than requiring the user to type responses freely. These bots are more like interactive forms than true conversational AI.'

Examples include;

Customer support chatbots

Chatbots when ordering food online

Booking appointments online

SO, just to clarify, these types of chatbots have:

Predefined paths: Users choose from pre-set options, and each button leads to another step or submenu.
No natural language processing (NLP) needed: The chatbot doesn’t “understand” your text — it just reacts to your clicks.
Decision trees: The flow is built like a tree with branches based on user choices.
Faster and safer: Great for simple tasks like booking appointments or FAQs, since users can’t go “off-script.”

So, they're just very very simple, personally wouldn't consider them to be 'real' AI

Generative AI Chatbots

2025-06-16 14:07:46 UTC

Examples include ChatGPT, Deepseek, MS CoPilot, customer service chatbots. These types of chatbots use AI to create new content (such as text, images, music and audio) based on large sets of data that it has been trained on.

Hybrid Chatbot

2025-06-16 14:07:47 UTC

Chatbot consisting of both 'rule-based responses' (pre-written FAQ's and knowledge) and AI chatbot interactions. So can answer basic questions and more complex situations using machine learning or AI and context-based on conversation to answer queries.

GPT vs AGENT

2025-06-16 14:29:08 UTC

Analogy:

GPT = Like a really smart librarian: you ask questions, it gives you answers.
Agent = Like a study tutor: it picks what you need to revise, quizzes you, gives you tasks, and adapts over time.

tbirdcymru — 2025-06-16 14:29:58 UTC

Platforms for chatbots:

ChatGPT Custom GPTs -

Amboss - USA

GitHub

Google - DialogFlow

MS Bot Framework

---------------------------

A GPT can be an AI agent, but not all AI agents are GPTs.

Difference between GPT and Agent -

A GPT becomes part of an AI agent when it's embedded in a system with sensors and actuators. It interacts with tools and APIS -- reads a user's calendar and schedules .