CMJ 103: Public Speaking

Artificial Intelligence (AI) is "the capability of computer systems or algorithms to imitate intelligent human behavior" (Merriam Webster). There are four basic components that are important to understand:

• Machine Learning - ability to combine data with algorithms
•      Examples - Facial recognition, targeted marketing
• Natural Language Processing - training computers to understand and process human language
•         Examples - Speech recognition, chatbots
• Deep Learning - uses additional layers of algorithms to recognize more complicated patterns and understand more complex processes
•         Examples - Image recognition, Google Translate
• Large Language Models (LLMs) - trained with huge amounts of textual data to recognize patterns, and generate output based on those patterns
•         Examples - ChatGPT, Claude, Perplexity

Generative AI (GenAI) combines all of the above concepts in tools that can put together content based on models, but not "create" content.

Very basically, Large Language Models (LLMs) are used with Natural Language Processing and Deep Learning to train GenAI tools to recognize patterns and use these patterns to respond to queries.

*** Before using any GenAI tools, be aware that there can be privacy concerns and anything you share with the tool may be saved and added to it's LLM***

Using some tools can provide helpful indications of where to focus your search for information.  Here are a couple that might be helpful:

• Claude - a tool by Anthropic, generally includes recognition when other sources might also be helpful to use
• Perplexity - a tool by a company of the same name, includes direct citations/links for the information it provides

It is important to remember that using GenAI tools for research is never the only solution, but they can provide helpful places to start, i.e. identifying one or two initial sources, indicating which disciplines are likely to cover what you need, etc.

There are many things to watch out for when using GenAI tools.  Here are three important ones:

• Misleading or Fabricated Information - GenAI tools do not "think," they only replicate patterns, thus when asking for a list of journal articles, for example, you will receive a list, but some of the citations could be completely made up, although they look legitimate.  You always need to double check information you receive from GenAI tools.
• Biases - Bias is not a new problem (it exists in all information).  But it continues with the LLMs used to develop GenAI tools; whatever biases are included in the data used to train the tool are replicated, and potentially amplified, by the GenAI tool's output.
• Gaps - Information received from GenAI tools is never comprehensive; there are always other tools that should be used, depending on what you need to find. For academic work, library databases are still vital!