Generative Artificial Intelligence (Generative AI, GenAI, or GAI) is a subset of AI that creates content such as text, images, videos, and other data types. It learns patterns and structures from training data, enabling it to ‘create’ new content.
Below are a few examples of Generative AI models that are relevant to studying at university:
Open AI – ChatGPT – Large Language Model (Conversational Understanding, Information Retrieval, Problem-Solving & Reasoning, Creative Writing, Coding Assistance)
DeepSeek – Large Language Model (Complex Reasoning, Mathematics, Coding)
Google AI – Gemini – Large Language Model (Successor to LaMDA, Conversational Understanding, Information Retrieval, Problem-Solving & Reasoning, Creative Writing)
Anthropic – Claude – Large Language Model (Text Generation, Conversational Understanding, Ethical Reasoning)
Microsoft - Microsoft CoPilot – Multimodal AI (Text Generation, Image Generation, Integration with Office 365 tools for productivity tasks like document creation, data visualisation, and summarisation)
Meta - Meta AI – Multimodal AI (Text Generation, Image Generation, Creative Assistance, Research Support)
Open AI - DALL.E 3 – Image Generation Model (Creative Image Design, Visual Content Creation, Artistic Rendering)
Padlet – Image Generation Model (Collaborative Visual Content Creation, Simple Design Assistance, Idea Visualisation)
Canva – Image Generation Model (Graphic Design Support, Visual Presentation Creation, Branding and Marketing Assistance)
Adobe - Adobe Firefly – Image Generation Model (Creative Visual Design, Advanced Photo Editing, Branding Assets Generation)
Generative AI models; as those mentioned above, are capable of a wide range of tasks, including generating text from user inputted responses. The responses of the models are generally unique, though can produce similar outcomes based on what they are trained upon.
Generative AI can be valuable for understanding concepts, producing idea prompts and understanding different perspectives as well as creating images in some image generating models (such as Adobe Firefly, Padlet, DALL.E, Canva). Using AI alongside your own critical thinking and creative problem-solving skills can assist you greatly within your degree.
DeepSeek focuses primarily on developing Large Language Models and was founded in 2023, with its first AI model being released the year after. It has been most popular for recently crashing the NVIDIA stock by £483 billion due to reportedly requiring a fraction of the cost to train and run than the US owned Open AI’s o1 model. DeepSeeks R1 proves itself to be a successful AI ‘Reasoning’ model (like OpenAI’s o1) and has similar capabilities.
As above, some AI programmes are able to produce images from written text prompts
Here is an example from the free image generator in Padlet from the prompt "A diverse group of university students play a board game in a beautiful library."
Have a look at the image closely. What do you see? Look at the individual features...
Images from generation software can appear decent at first glance, however due to restrictions they have problems replicating human faces and other features too. Complex images are likely to possess many errors, a few from above are:
In conclusion, you can use AI images for simple illustrations, but sometimes it is easier to do things manually or find creative commons licensed images instead.
AI can also be used in image editing. Adobes Firefly & Photoshop and Google's 'magic eraser' tools can be used for AI assisted image editing. As shown above, the AI is able to recognise the obstructions within the photograph and 'suggest' the removal of such. It uses Machine Learning to predict what the image 'should' look like, and fills it in based on its findings.
AI tools like Consensus have practical applications in locating and summarising research information. Users can pose research questions to Consensus, which responds with quotes from academic sources that may answer the query.
Consensus also features a "synthesise" function, summarising findings from accessible papers and presenting them in a Consensus Meter, which indicates the level of agreement among sources. Users can click on individual sources to view their abstract, citation count, and options for accessing the full text.
However, it is important to note the limitations of Consensus. The tool only analyses a handful of papers at a time, which can potentially lead you to miss critical information. Understanding the relationships between texts and the significance of certain facts cannot be fully achieved through this model. Therefore, it remains essential to read the texts yourself to assess their relevance and appropriateness. Consensus also advises reviewing the papers it identifies and consulting additional sources for a more thorough understanding.
Example: We asked Consensus, "Can mindfulness improve sleep?"
Consensus identified 16+ "highly relevant results" and analysed 13 papers to populate the Consensus Meter.
The Consensus Meter itself is not 100% accurate, so its findings should be taken with a pinch of salt.
AI tools; like Consensus, can provide an initial insight into a research topic but should not replace your own research skills. It is advisable to verify the AI’s findings by reviewing the identified sources and conducting independent research.
Berrada, D. (2023). Magic Eraser plus More Google Photos Features Coming to Google One. [online] Google. Available at: https://blog.google/products/photos/magic-eraser-android-ios-google-one/
Dartmouth (1956). Artificial Intelligence (AI) Coined at Dartmouth | Dartmouth. [online] home.dartmouth.edu. Available at: https://home.dartmouth.edu/about/artificial-intelligence-ai-coined-dartmouth.
Google (2023). Magic Eraser plus More Google Photos Features Coming to Google One. [Digital Image] Google . Available at: https://blog.google/products/photos/magic-eraser-android-ios-google-one/.
Ithaka S+R. (2024). Generative AI Product Tracker - Ithaka S+R. [online] Available at: https://sr.ithaka.org/our-work/generative-ai-product-tracker/.
Padlet (2022) A diverse group of university students play a board game in a beautiful library [Digital Image].
