What is Generative AI?
Generative AI refers to a type of artificial intelligence designed to create new content or data that resembles existing patterns or examples. This type of AI uses massive databases to take existing content, and transform it to answer new queries. This largely began with Chat GPT 3, and we are already onto GPT-4, with GPT-5 on the horizon.
For instance:
- Text: Models like GPT-4 can generate coherent and contextually relevant text based on a given prompt. They can write essays, create dialogue, or even generate creative stories.
- Images: Generative models like DALL-E or stable diffusion can create images from textual descriptions, combine visual elements in new ways, or even generate entirely new visual concepts.
- Music: AI can compose original pieces of music or generate variations on existing tunes.
- Code: Some models can write code snippets or even whole programs based on functional requirements or example inputs.
Benefits and Drawbacks of AI
Benefits
The benefits can be seen when we look at the use of AI in business. Generative AI has the potential to increase the productivity of each individual employee, both by doing their job more efficiently, and by being able to do jobs they otherwise wouldn’t be able to, due to the assistance of AI.
The use of AI in education can be helpful in certain learning experiences. Here’s a list of some ways it could be helpful for you and your students:
- It can assist in the creation of customized learning experiences for students.
- It can help students get started with an assignment or research papers.
- It can help faculty identify gaps in curriculum.
- It can be helpful to determine an alternative way to write a paragraph- maybe in a more friendly, helpful tone.
- AI can be useful for (some) tutoring and tasks sometimes handled by a TA.
Drawbacks
The bad news is that the current tools are not good at differentiating between human and AI inputs, often giving false positives. Here’s a list of some known concerns with AI:
- The AI experience can be customized for the learner, resulting in the response being customized to the learner. Each learner may get a different response to a similar question. A student who is working to improve vocabulary skills will get a different response than one who uses advanced vocabulary already. This can create inequalities among learners.
- Many AI tools “collect information”. The systems use this information to offer a more polished response to the next user. A best practice is to refrain from mentioning identifying information like “SUNY Oswego” or your name to maintain privacy and security.
- AI tools are known to have biases. There are conversations about “machine-learned biases” among other biases. This bias, in particular, has to do with the user providing multiple prompts indicating a certain bias. The machine catches on to that bias so now the responses are biased, too.
- Some AI tools assist with image generation. The images generated tend to demonstrate stereotypes, which could misrepresent a group or country.
- There are concerns about misinformation and not offering reliable sources of information.There are times when references are “made up” or don’t exist.
Prompt Engineering
Prompt Engineering is the concept of providing the AI tool with a quality prompt so that the response given is more closely related to what is being looked for. Ultimately, if you want a quality response from AI, you need to provide a quality prompt.
When asking AI for information, the quality of the response depends on the context provided in your prompt. Prompt engineering helps to reduce generalized responses. When creating a prompt, consider including the following information:
- Task (what you are asking AI for assistance with)
- Context/background (intended audience, scenario of where the response will be used)
- Persona (information about you)
- Tone (friendly, kind, helpful)
Here is an example:
You would like to know what prompt engineering is and how it impacts your role on campus. Consider the difference between the two prompts listed below:
General prompt: What is prompt engineering?
Quality prompt: I am a professor who has never used AI in the classroom. I have some computer technology skills. I’d like to know what prompt engineering is and how I can use it in my online course. Please be kind with your response and use plain language.
The difference is that the quality prompts include a task, some context/background information, some broad details about the instructor, followed by asking the question and providing the tone it should be shared.