Workshops

The information world of today is complex and deep, filled with many different kinds of publications and online content types, as well as untrustworthy junk such as 

• misinformation and disinformation 
• propaganda 
• AI slop (for a fun take on AI slop, also watch this John Oliver video; contains NSFW content!).

All this makes it hard to filter out or separate unreliable from reliable information.

The ability to effectively evaluate the trustworthiness of information sources for a specific task is a foundational skill for academic success and also one highly valued by employers. The videos and accompanying text and practice exercises below will help you build and reinforce these skills, which can then be honed further by using them in your coursework.

As you work through the content below, here are some questions to consider:

• What factors can you look at to help determine a source's credibility (trustworthiness)?
• What is lateral reading, and how is it used to check a source's credibility?
• Can evaluation strategies also be used for generative AI outputs? Are there factors specific to AI that should be considered?

Critically evaluating sources used in academic work (alongside citing them properly) is an important aspect of ethical use of information, which is itself a part of the University Learning Goals and the SUNY General Education Information Literacy core competency. 

Evaluation of sources applies to open web sources and those found in libraries, but the level of scrutiny generally should be higher for web sources. For example, peer-reviewed journal articles (often synonymous with scholarly articles) and often found in the library have already gone through a peer-review publishing process designed to filter out most erroneous or misleading content.

Here are two ways to help you determine the credibility of a source:

1. Look at specific factors about the source. The method we'll focus on is the CRAAP Test. Watch the videos below and look at the CRAAP Test handout for more details.
2. Fact-check a source by reading laterally. Find and read other sources to make sure you can verify facts stated in the original source. The second video below explains this process in more detail. With current 'news' there are also fact-checking organizations that investigate.

Generative artificial intelligence (genAI) tools are those that generate content using predictive analysis based on vast amounts of human-generated training data and user prompts. Standalone products such as Open AI's ChatGPT, Microsoft's Copilot, or Google's Gemini sit alongside similar tools integrated into existing products and applications. This includes some library databases like OneSearch and Scopus and web and phone apps such as Grammarly, search engines, and more.

Such applications can be useful if you know where they are getting their data from (library database AI tools only use data from the content indexed in that tool, rather than from the open web) and how to use them effectively.

For your course work check your course syllabus or learning contract (or ask your instructor) for restrictions or explicitly allowed uses of genAI. These limits can differ from course to course. When allowed, properly attributing and citing where and how you used gen AI in your coursework is required to meet University academic honesty policies.

Also be aware that genAI has issues with providing incorrect information and fabricated sources (you will often see this latter behavior referred to as "AI hallucinations"). Fact-check any information gained from such tools. Lateral reading is an excellent strategy to do this.

The video below is a brief introductory discussion of some of the evaluation issues to consider when using generative AI tools.

1. Search the library and the open web for 2+ sources on a research topic of interest to you. Spend a couple of minutes applying the CRAAP test to those sources. What does that tell you about the reliability of those sources?
2. Also, explore an "AI assistant" tool within a library database on your topic. Try either OneSearch (after you do a search, on the results page, click the 'Research Assistant' logo on the far right of the screen) or Scopus (on the landing page, click 'Scopus AI' link above the search box). In both these cases, the generative AI tool is working only from the data within the databases; i.e., mostly scholarly journal articles. If you compare that to how popular tools like ChatGPT function (trained on vast amounts of internet-scraped information with little ability to differentiate reliable from unreliable information), you'll see a difference in the quality of their output.
3. (for fun) Take the AI Video quiz (via New York Times). Can you tell AI-generated content from real photos/videos?

If you didn't read or view the materials above, please do so now:

Work through the content above and take this quiz:

• Source Evaluation quiz (it includes a Certificate of Completion after submitting that can be emailed, downloaded, or printed out)

Note that without carefully reading the content above and watching the videos you will probably not be able to answer many of the questions.