Capabilities of Artificial Intelligence refer to the tasks and functions that AI systems can performespecially those that typically require human intelligence. These capabilities are what make AI valuable in areas like education, healthcare, business, and libraries
Limitations of Artificial Intelligence are the challenges, weaknesses, or boundaries that restrict what AI can do effectively. While AI is powerful, it is not perfect and often struggles with tasks requiring human judgment, creativity, ethics, or adaptability.
What AI Can Do
Automation: AI handles routine tasks like email alerts and FAQs, boosting efficiency.
Data Analysis: AI quickly processes large data sets to improve resource use and engagement.
Predictive Analytics: AI forecasts trends like study room demand or resource usage.
NLP: AI understands and responds to user queries, aiding communication and translations.
Computer Vision: AI processes images, e.g., digitizing manuscripts or scanning IDs.
Personalization: AI offers tailored suggestions based on user preferences.
24/7 Support: AI chatbots provide instant help anytime for better accessibility.
What AI Cannot Do ❌
Bias and Discrimination: AI can inherit biases from training data, risking skewed results. Always validate AI-generated content with credible sources to avoid perpetuating stereotypes.
Plagiarism Risks: AI content may closely resemble existing work, raising plagiarism concerns. Researchers must ensure originality and uphold integrity.
Data Misinformation: AI may produce inaccurate information. Cross-check outputs with reliable sources to prevent spreading false data.
Hallucinations: Tools like ChatGPT or Microsoft CoPilot can fabricate plausible but false citations or information, known as “hallucinations.”
Paywalled Content: AI literature search tools lack access to paywalled articles. They assist in searching but cannot replace human access to full texts.
Training Data Limits: AI outputs depend on its training data scope and cut-off dates. For example, ChatGPT-4’s knowledge only goes up to April 2023.
Reproducibility: AI-generated content is non-reproducible; different users may get varied results from the same prompts at the same time.
Ethics, Privacy, etc.: There are numerous limitations related to ethics, privacy, bias, labor and environmental impact outlined .
Benefits of Generative AI Tools
Brainstorm: Tools like Microsoft CoPilot* and Gemini help generate ideas, organize thoughts, and overcome writer's block.
Break down concepts: These tools assist in understanding complex concepts or assignment prompts.
Illustrate: Image generation tools like Midjourney and Dall-e aid in illustrating work.
Create: AI tools support creative adaptations.
Summarize: Tools like CoPilot and ResearchRabbit can summarize articles efficiently.
Discover research: AI tools like ResearchRabbit and Elicit help discover and visualize new research.
Translate: AI tools facilitate language translation.
Code: GitHub Copilot helps generate and refine code.
(*Microsoft CoPilot is licensed by UT
AI Future Directions
Strategic Foresight and Governance: Governments must prioritise AI safety, set liability rules, and invest in transparent, accountable development.
Balanced Innovation: AI advancement should serve public good—healthcare, climate, inclusion—while embedding ethics and fairness.
Workforce Transformation: Reskilling programs and education must support workers adapting to AI-integrated environments.
Enhanced Collaboration: A global, inclusive approach involving civil society, NGOs, and nations is essential for managing AI’s reach.
Continued Technological Growth: AI will continue to drive innovations that reshape economies and societies, if risks are proactively managed.
Robust Cybersecurity Measures: Embedding security in AI systems is vital to protect assets, privacy, and trust in digital infrastructure.
AI’s potential futures: Mitigating risks, harnessing opportunities ...