Information management Exam 2 questions and answers

Why do large language models sometimes “hallucinate”?

LLMs hallucinate because they generate text by predicting likely word patterns rather than checking facts. If they lack reliable information, they may still produce a confident answer that sounds correct even when it is false.

Give three reasons why the output of large language models can contain factual errors.

First, training data from the internet can contain incorrect or outdated information. Second, models predict words instead of verifying facts. Third, compressing huge datasets into patterns can cause details to be mixed up.

Why do large language models often struggle to answer questions involving even simple logic?

LLMs are optimized for predicting language patterns, not strict rule-based reasoning. Logical problems require step-by-step consistency, which probabilistic language prediction does not always maintain.

Explain why unassisted large language models can generate coherent stories a few pages long but not full-length novels.

They can maintain context for a limited amount of text, allowing short stories to stay coherent. However, long novels require long-term planning and consistency, which models often lose as the text grows longer.

Explain why machine learning has largely displaced expert systems.

Expert systems required humans to manually write and maintain large rule sets. Machine learning can automatically learn patterns from data, making it more flexible and easier to scale.

Describe three negative consequences of using AI to complete a programming assignment.

The student may not learn the programming concepts, may struggle on exams that require understanding, and may violate academic integrity policies if the assignment requires original work.

How would you ensure AI-generated memes for NFTs do not infringe on copyrights?

Use original or public-domain content and avoid recognizable copyrighted characters or logos. Also check the AI tool’s license to ensure commercial use is allowed.

Explain why large language models are built using neural networks instead of regression models.

Neural networks can learn complex patterns in massive datasets such as language or images. Regression models are simpler and cannot capture these complicated relationships effectively.

Why are copyright laws relevant to artificial intelligence?

AI models are trained on large datasets that may include copyrighted text or images. Copyright laws determine whether this use is legal and how AI-generated content can be shared or sold.

Explain why AI training data is generally considered covered by fair use principles.

Training is often considered fair use because the model learns patterns rather than reproducing the original content. The data is transformed into statistical representations instead of being copied directly.

Explain the difference between trademarks, copyrights, patents, and trade secrets.

Trademarks protect brand names and logos. Copyrights protect creative works like books and music. Patents protect inventions and technologies. Trade secrets protect confidential business information kept private.

Why are current AI models dependent on GPU hardware?

Training neural networks requires huge numbers of mathematical calculations. GPUs perform many calculations at once through parallel processing, making them much faster than CPUs for AI tasks.

What challenges do educators face following the widespread availability of AI tools?

Educators must ensure students are learning instead of relying on AI for assignments. They may need new assessment methods and updated academic integrity policies.

Will AI be a net benefit for society?

AI can increase productivity, automate repetitive tasks, and help solve complex problems. However, society must manage risks such as job displacement and ethical concerns.

Give three examples where businesses might use AI agents.

Businesses can use AI agents for customer service chatbots, automated scheduling or administrative tasks, and analyzing company data to generate reports.

Does AI understand the board games it plays well?

No, AI does not truly understand the game. It analyzes patterns and calculates optimal moves using algorithms rather than having human-like awareness or understanding.

Why is playing chess easier for AI than image recognition?

Chess has clear rules and limited possible moves that can be represented mathematically. Image recognition is harder because images contain complex visual information that must be interpreted.

List three advantages of regression analysis over machine learning.

Regression models are easier to interpret, require smaller datasets, and help researchers understand cause-and-effect relationships between variables.

Explain what labeled and unlabeled data mean.

Labeled data includes inputs with the correct answers, such as emails marked spam or not spam. Unlabeled data contains only inputs, so the model must discover patterns without known outcomes.

How does machine learning extend the predictive powers of linear regression?

Machine learning can model complex and nonlinear relationships between many variables. This allows more accurate predictions on large and complicated datasets.

Give three examples of demonstrating soft skills in a job interview.

Clearly explaining how you solved a problem shows communication. Describing teamwork on a project shows collaboration. Explaining how you learned a new skill shows adaptability.

How will entry-level work in finance and accounting be affected by AI?

AI will automate routine tasks like data entry and basic reporting. Entry-level workers will focus more on interpreting data and supporting business decisions.

What entry-level business work is unlikely to be replaced by AI soon?

Roles requiring human interaction such as sales, consulting, and management training are less likely to be replaced because they rely on relationship building and judgment.

How will employer hiring expectations likely change as AI becomes common?

Employers will expect workers to know how to use AI tools and analyze data. They will also value creativity, communication, and critical thinking more.

List four human skills likely to decline due to AI adoption.

Manual data entry, basic information lookup, routine report writing, and repetitive administrative tasks may decline because AI can automate these activities.