Skip to Main Content

Artificial Intelligence: AI Tools

General information and guidance on the use of AI for the UCD community
Midjourney generated caricature of AI as a smart, biased, supremely confident, drunk intern.
Midjourney generated caricature of AI as a smart, biased, supremely confident, drunk intern.4

 

There is a continuous stream of publicity around the various GenAI tools, with newer, better, faster models entering the mix seemingly every other week. The hype assures us that these new or improved models have eliminated previous issues and that their reliability is assured. However, notwithstanding the noticeable and quantifiable improvements as the evolution of GenAI models progresses, when considering these assertions it is best to engage critical thinking skills when assessing the outputs of GenAI tools.

Below you will find a curated selection of GenAI tools organised by the type of output they generate. When engaging with GenAI tools it is advisable to do so judiciously, to carefully review their outputs critically to ensure the quality of the content before incorporating it into your work and to be entirely transparent about where and how they were used. You should also be sure to review the privacy and data protection and use policies before submitting your personal data to AI tools.

In July 2023, Punya Mishra likened GenAI reasoning models to "a smart, biased, supremely confident, drunk intern".1 In February 2025, after repeating his assessment with newer models, Mishra concedes that the newer models perform much better than their predecessors when assigned certain tasks "like having that intern suddenly sober up, pivot into genius mode, and deliver the perfect solution right when it was needed."2 However, this improvement is not consistent across all types of tasks leading him to conclude that, "It’s witty, but it also rings true: you get these flashes of brilliance mixed with the occasional wild misstep."3

UCD IT Services have a list of AI services that are available which have been verified for compliance with the university's privacy and IT security guidelines. Free or alternative services may not have the same protections in place.

You can find additional AI tools through directories such as AIxploria, Futurepedia, and AI Tool Library amongst others. UCD Library provides these lists for information purposes and their inclusion here should not be inferred as an endorsement.

References (Vancouver):
  1. Mishra, Punya "ChatGPT is a smart, drunk intern: 3 examples"  Arizona, 26 July 2023. Available from: https://punyamishra.com/2023/07/26/chatgpt-is-a-smart-drunk-intern-3-examples// (Accessed 25 February)
  2. Mishra, Punya "GenAI Reasoning Models: Very smart & confident (but still drunk)"  Arizona, 14 February 2025. Available from: https://punyamishra.com/2025/02/14/genai-reasoning-models-very-smart-confident-but-still-drunk-intern/ (Accessed 25 February)
  3. ibid.
  4. Midjourney v. 6.1. Response to "Caricature of AI as a smart, biased, supremely confident, drunk intern" AI generated image. https://cdn.midjourney.com/413f85c3-0b80-426c-b6fa-08f9ce7572f1/0_2.png Midjourney Inc.,  14 March 2025.

AI Tools, by type

Generative AI Tools for Text Generation: A curated list

 

Gemini (Google): Offers real-time information access through Google's search integration, making it valuable for current research trends and recent publications. Strong in scientific and technical domains, with capabilities in data analysis and visualization. Particularly useful for literature discovery and staying current with research developments. Main limitations include less consistent performance compared to GPT-4 and Claude in complex reasoning tasks, and occasional challenges with maintaining academic writing style.
Access: Available through UCD IT Services Campus licensing.

NotebookLM (Google): NotebookLM is a tool that helps you understand complex information by summarising sources and providing relevant quotes. Now, you can listen to AI-generated discussions about your sources with one click, using the new Audio Overview feature.
Access: Free with a Google account login.

Claude (Anthropic): A versatile language model particularly suited for academic work due to its strong capabilities in analysis, technical writing, and citation handling. Excels at complex reasoning tasks, literature review assistance, and methodological discussions. Notable for maintaining academic rigour and detecting potential biases in research design. Primary limitations include potential for hallucination, particularly with specific citations, and a knowledge cutoff date that requires verification of recent information. Best used for brainstorming research questions, refining methodological approaches, and analyzing research designs.
Access: Free, Pro, Team and Enterprise plans available.

ChatGPT (OpenAI): Advanced language model with strong capabilities in code generation, mathematical problem-solving, and interdisciplinary analysis. Particularly useful for research proposal development, statistical analysis planning, and technical writing review. Demonstrates sophisticated understanding of academic contexts and can assist with grant writing and methodology development. Known limitations include inconsistent citation accuracy, potential for confident but incorrect responses, and challenges with maintaining context in longer academic discussions. Requires careful verification of outputs, especially for technical content.
Access: Free, Plus and Pro pricing plans available. 

CoPilot (Microsoft): Available through UCD's Microsoft Office 365 subscription, with EU security and compliance requirements built-in. Stand alone chat interface, but also integrated within all Microsoft Office applications through enhance features such as PowerPoint Designer. 
Access: Free and Pro pricing plans.

Perplexity AI: Combines LLM capabilities with real-time web search, making it particularly valuable for current literature reviews and identifying recent research trends. Provides citations for its sources, though these should be independently verified. Limited in   complex mathematical concepts and may sometimes misinterpret technical papers. Best used as a starting point for literature exploration rather than comprehensive analysis.
Access: Free, Pro and Enterprise pricing plans.

Generative AI Tools for Image Creation: A Curated List

 

DALL-E 3 (OpenAI)
An image generation model that creates highly detailed scientific illustrations and technical diagrams. Useful for visualizing complex concepts, creating custom teaching materials, and generating accurate anatomical or molecular representations. Notable for its ability to handle technical and scientific terminology accurately. However, users should note it occasionally struggles with precise mathematical notation and may produce inconsistencies in specialized scientific symbols.
Access: Via subscription to ChatGPT Plus and Enterprise services.

Midjourney
A good source of visualizations for architecture, design, and biological sciences. Creates detailed structural and organic forms, making it valuable for conceptual architectural models, biological illustrations, and visual abstracts for publications. The primary limitation is its closed-source nature and the requirement to use Discord, which may present institutional access challenges. Also lacks direct integration with academic workflows.
Access: No free access. Subscription to Basic, Standard, Pro and mega plans available.

Stable Diffusion (Open Source)
An open-source platform that allows full control over the generation process and can be deployed locally, making it ideal for research requiring data privacy or custom training. Particularly valuable for specialized research fields as it can be fine-tuned with domain-specific datasets. The main drawbacks are the technical expertise required for setup and the computational resources needed for local deployment.
Access: Free (Open Source).

Adobe Firefly
Designed for creative and professional use, with strong integration into the Adobe Creative Suite. Generates and edits images within academic publishing workflows and creates consistent visual styles for educational materials. Its key advantage is the focus on commercially safe content generation, making it suitable for academic publications. The main limitation is its subscription cost and relatively smaller range of academic-specific features.
Access: Standard, Pro and Premium plans available.

Leonardo.ai
Offers specialized features for scientific visualization and technical illustration, with strong capabilities in generating consistent series of related images. Useful for creating educational materials and research presentations. Includes features for maintaining visual consistency across multiple generations, which is valuable for creating comprehensive teaching materials. However, it has a steeper learning curve compared to other tools and may require more detailed prompting for accurate scientific content.
Access: Free plan [with limited token-based access to image generation tools], Apprentice, Artisan, maestro and Enterprise plans available.

A crucial note for academic users: All AI-generated images should be clearly labeled as such in academic work, and institutions may have specific policies regarding their use in formal publications. It's recommended to check with relevant journals and institutions regarding their policies on AI-generated images before incorporation into academic work.

 

Generative AI Tools for Image Creation: A Curated List

 

Runway Gen-2
A text-to-video generation platform  for creating research visualizations and concept demonstrations.Produces short (3-4 second) high-quality clips that can enhance academic presentations and teaching materials. The platform offers good control over visual style and motion, making it valuable for disciplines requiring visual explanations of abstract concepts. Main limitations include clip length restrictions and occasional artifacts in complex motion sequences.

Pika Labs
Specifically designed for scientific and technical content creation, offering capabilities to generate accurate representations of scientific processes and phenomena. Notable for its ability to maintain scientific accuracy in visualizations, making it particularly useful for STEM fields. The tool includes features for adding scientific annotations and maintaining visual consistency across generated sequences. Primary drawbacks include limited customization options for specialized scientific contexts and occasional issues with complex technical details.

DeepBrain AI
Focuses on creating AI avatars for educational content, particularly useful for developing lecture materials and research presentations. Offers multilingual support and maintains consistent presenter appearance across multiple videos, beneficial for creating comprehensive course materials. The platform includes features for academic-specific content like citation displays and technical terminology pronunciation. Main limitations include occasional unnatural movements and limited emotional range in AI presenters.

Play.ht

A text-to-speech generator with a variety of AI voices to choose from, including a voice cloning option. Useful for creating e-learning materials with accessible audio formats that can be easily updated. The tool includes Wordpress plugins and Chrome extensions that make it easy to integrate into a users' workflows. Limitations include the restrictions of the free plan, and a narrow selection of voices for non-English languages.

D-ID
Specialized in creating presenter-led educational content. Features include support for multiple languages, integration of academic slides, and maintaining professional presentation standards. Useful for creating reusable lecture content and research presentations. Main limitations include restricted customization of academic presentation formats and occasional issues with technical terminology pronunciation.

 

Important Note: As these tools continue to evolve rapidly, academics should regularly verify their outputs against current research standards and institutional guidelines for academic integrity. It's recommended to maintain transparent documentation of AI tool usage in academic work and to verify all generated content for accuracy.

Generative AI Tools for Presentation Slides Generation: A Curated List

 

Beautiful.ai:

  • AI-powered presentation software that automatically adjusts layouts
  • Maintains design consistency as you add content
  • Includes smart templates that adapt to your content
  • Real-time collaboration features
  • Built-in asset library
  • Main limitation: Mostly aimed at a corporate market, its educational options are minimal.

Canva Presentations with Magic Design:

  • AI features help with layout and design choices
  • Can generate presentations from outlines
  • Extensive template library
  • Includes AI image generation and editing tools
  • Easy-to-use interface for non-designers
  • Main limitation: Slide outputs are quite basic and not compatible with other presentation applications

Gamma:

  • Creates presentations from text prompts
  • Generates professional-looking slides with minimal input
  • Can import content from various sources including docs and websites
  • Strong data visualization capabilities
  • Includes AI-powered image generation
  • Main limitation: Very restricted free plan and lack of offline editing even with paid plans

Microsoft 365 Copilot for PowerPoint:

  • Integrates directly into PowerPoint
  • Suggests design layouts and formatting based on your content
  • Can generate entire presentations from simple prompts or outlines
  • Helps create consistent styling across slides
  • Can suggest relevant images and icons
  • Main limitation: Slide designs may lack in creativity, and copilot can be prone to hallucinations.

SlidesAI.io

  • Specializes in converting complex academic content into structured learning materials.
  • Particularly useful for creating teaching materials and knowledge-building frameworks. 
  • Helps break down complex research papers and theories into digestible learning modules. 
  • Main limitation: its occasional oversimplification of complex academic concepts and some difficulty with highly technical content.

Since this is a rapidly evolving field, it is recommended that you to check the latest features and pricing for any tool you're interested in. 

Generative AI Tools for Code Generation and Analysis: A Curated List

 

GitHub Copilot: Assists in research code development, particularly useful for data analysis scripts and research software development. Can significantly speed up coding tasks but may suggest incorrect or inefficient solutions for complex scientific computing problems. Requires careful review and testing of generated code.
Access: Free, Pro, Business and Enterprise plans available.

TabNine: Provides context-aware code completions for research programming tasks. Particularly useful for maintaining consistent coding style in research projects. May sometimes suggest completions that don't align with scientific computing best practices and requires careful validation.
Access: Preview (14 day trial), Dev and Enterprise plans available.

CodeT5+ (Google): CodeT5 is a pre-trained encoder-decoder model that uses identifier-aware objectives to learn from code tokens. It can perform various tasks such as code generation, translation, repair, and vulnerability detection.
Access: Open Source (free)

Code Llama (Meta): An open source model specifically trained for code generation. Code Llama is a code-specialized version of Llama 2 that was created by further training Llama 2 on its code-specific datasets, sampling more data from that same dataset for longer. Three models are available: 1. Code Llama; 2. Code Llama Python; 3. Code Llama Instruct.
Access: Open Source (free)

Generative AI Tools for Academic Research: A Curated List

 

Searching tools

Semantic Scholar: AI-powered academic search engine that excels at finding relevant research papers and understanding conceptual connections between studies. Offers sophisticated citation analysis and research impact metrics. Particularly useful for understanding research landscapes and identifying key papers in a field. Limited by database coverage and occasional challenges with very recent publications. Best used in combination with traditional literature search methods.
Access: Free

Research Rabbit: Visualize networks of papers and co-authorships. Use graphs as new "jumping off points" to dive even deeper! Collaborate on collections, or help kickstart someone's search process! And leave comments as well!
Access: Free 

 
Research Writing and Editing Tools

Writefull: Specifically designed for academic writing, offering language enhancement and style suggestions tailored to scientific publications. Particularly useful for non-native English speakers in academic publishing. While effective for general academic writing, it may sometimes suggest overly formal alternatives and struggle with highly technical or field-specific terminology.
Access: Free, Premium and Institutional pricing plans (15% student discount)

Scite.ai: Uses AI to analyze citation contexts and provide insight into how papers are cited in the literature. Helps researchers understand the impact and reception of publications. Particularly valuable for literature reviews and understanding research context. Limited by database coverage and may miss important citations from newer or less mainstream publications.
Access: 7-day free trial, Personal and Organisaton pricing plans.

 

Data Analysis and Visualisation

Zams (formerly Obviosuly AI): Automates complex data analysis tasks and generates interpretable insights from research data. Particularly useful for exploratory data analysis and hypothesis generation. While powerful, it may oversimplify complex relationships and should not replace thorough statistical analysis for publication-ready research.
Access: Startup, SMB and Enterprize pricing plans

Elicit: AI enabled Systematic Review. Specializing in research question exploration and literature analysis. Particularly strong in identifying relevant papers and summarizing research findings. However, its coverage can be limited for very recent publications or niche fields, and its summaries should be verified against original sources.
Access: Free, Plus, Pro, Team and Enterprize pricing plans

Generative AI Tutors/Knowledge Builders: A Curated List

 

AI Tutor Pro (Contact North/Contact Nord): AI Tutor Pro is an open source Generative AI student study tool developed by Contact North | Contact Nord. This student-facing tool is designed to be a study aid for students. Featuring two distinct streams of interaction for knowledge building: Grow my knowledge and Check my knowledge.
Access: free

Tutor AI (Lumination AI): Tutor AI is a platform that offers custom courses and online courses to help you achieve more in school, work, and life. You can ask questions, get explanations, examples, and learn at your own pace with a personal AI tutor.
Access: Free, Mentor and Genius plans.

Julius (Caesar Labs): Julius is a powerful AI data analyst that helps you analyze and visualize your data. Chat with your data, create graphs, build forecasting models, and more. Get accurate help with the most proven AI Tutor.
Access: Free, Basic, Essential  and Pro pricing plans. (50% discount for students and academics)

Llama tutor (Meta): Enter a topic you want to learn about along with the education level you want to be taught at and generate a personalized tutor tailored to you! 
Access: Open Source (free)

 

Important Considerations:
  1. These tools should be used as supplements to, not replacements for, traditional academic mentoring and supervision.
  2. The effectiveness of these tools often depends on the quality of the input and the clarity of learning objectives.
  3. Regular validation of acquired knowledge through peer review and expert consultation remains essential.
  4. Users should be aware of any College, School or module specific policies regarding AI use in academic learning and development.