Skip to Main Content

Libraries & Cultural Resources

Research Guides

Generative AI Tools for Students

Generative AI is transforming how research is conducted, from literature reviews to data analysis and writing. While it offers powerful tools for productivity and creativity, it also raises important questions about authorship, integrity, and transparency

What is Generative AI?

Generative AI is a type of artificial intelligence that can create new content, such as text, images, music, or code, based on patterns it has learned from existing data. Unlike traditional AI, which mainly recognizes or categorizes information, generative AI can produce original outputs, making it a powerful tool for creativity, communication, and innovation across many fields.

To learn more about how generative AI works and why it's transforming education, business, and research, check out this short explainer video by KI-Campus:

How does it connect to the real-world

Generative AI is already making a real-world impact across many sectors:

Generative AI isn’t just a trend; it’s a growing tool shaping how we learn, work, and innovate.

Using AI Ethically & Transparently

 

As AI becomes more integrated into daily life, ensuring its ethical and transparent use is crucial. Ethical AI means developing and using AI systems responsibly, while transparency ensures that users understand how AI makes decisions.

Key Principles of Ethical AI

  • Fairness: AI should avoid bias and treat all users equitably.
  • Accountability: Developers and organizations must take responsibility for AI outcomes.
  • Privacy & Security: AI systems should protect user data and prevent unauthorized access.
  • Transparency: AI models should be explainable, allowing users to understand how decisions are made.

Why Transparency Matters

Transparency in AI helps build trust and ensures that AI systems are used responsibly. It involves:

  • Clear explanations of how AI models work.
  • Open discussions about AI limitations and biases.
  • Providing users with control over AI-generated content.

Sources & Further Reading

NOTE: Ucalgary students need to review their course outlines to understand the instructor's expectations around Generative AI use in their course. Every instructor's approach may be different. Should you have questions about what you can and can not do with generative AI in a class the first step it to ask your instructor or TA to clarify their expectations.

What About Accuracy? Understanding Hallucinations

Hallucinations in AI refer to instances where models generate information that appears plausible but is actually incorrect or fabricated. This phenomenon can significantly impact the accuracy of AI outputs.

Causes of Hallucinations

  • Incomplete or Biased Data: AI models trained on incomplete or biased datasets may produce inaccurate results [1].
  • Complex Reasoning Processes: Advanced reasoning models, like OpenAI's o3 and o4-mini, tend to hallucinate more due to their intricate step-by-step reasoning [2].
  • Training Flaws: Issues in the training process can lead to higher rates of hallucinations [2].

Impact on AI Models

  • GPT-4.5: Has a lower hallucination rate compared to o3 and o4-mini models [2].
  • OpenAI o3: Hallucinates 33% of the time during PersonQA tests and 51% during SimpleQA tests [2].
  • OpenAI o4-mini: Hallucinates 41% of the time during PersonQA tests and 79% during SimpleQA tests [2].

Addressing Hallucinations

Efforts to reduce hallucinations include improving training data quality, refining model architectures, and implementing better validation techniques [1].

Understanding and mitigating hallucinations is crucial for enhancing the reliability and accuracy of AI models.

[1] When AI Gets It Wrong: Addressing AI Hallucinations and Bias

[2] Why AI ‘Hallucinations’ Are Worse Than Ever

Libraries & Cultural Resources