When prompting AI like ChatGPT to generate content, you are responsible for applying the program conscientiously.
A few weeks ago, I co-hosted a webinar about leveraging AI for AEC marketing. The webinar highlighted some things our marketing team has learned over the past few months as we’ve worked to implement ChatGPT in our go-to internal resources. We got more than 60 questions on the webinar. One particular question stood out to me, and my collegiate background in philosophy and ethics was stirred.
The question was, “What are the ethical conundrums with using content created by AI?”
This one is a heavy topic, but a necessary conversation to have as we navigate the new and exciting world of AI resources in the AEC industry.
I’ll start with a bit of background on AI. ChatGPT and other AI language models pull information from their own internal sources. Their sources originally learned things from the internet, social media, books, news outlets, and past user prompts.
When you type into ChatGPT’s log, it surfs its knowledge based on keywords in your prompt, analyzes content scenarios, and optimizes an output for you (the user). Each time you type in ChatGPT’s log, it also uses your words to build its knowledge base for future prompts.
This means that any potential content solution that AI generates is susceptible to dispute since everything that lives on the internet or was put into an AI model was created by real people. Those real people have biases, prejudices, and even access to sensitive information.
And while ChatGPT and other language models were written with the human sentiment in mind (that’s another discussion), they can sell you on a response without it even being accurate, because something else they lack is the ability to tell absolute difference between fact and opinion.
More simply put:
- ChatGPT can create content that was influenced by online bias and discrimination.
- ChatGPT may be able to reproduce sensitive content that may put others at risk.
- ChatGPT can’t always deduce the difference between right and wrong or fact and opinion.
That’s all kind of intimidating. And it also means that we can’t blame AI for reproducing content that doesn’t align with our values.
We should be conscientious about this because we can’t trust that other people using AI to generate and publish content are doing so with similar values in mind. Which leads to more content living online that hasn’t been vetted for accuracy – which our AI counterparts can then pull information from.
So, when prompting AI to generate content, you (the user) are responsible for applying the program in keeping with its potential flaws and knowing that other humans are very well part of the circular ethical conundrum that AI provokes.
If you choose to use AI for brainstorming or content creation, please use AI conscientiously by doing two things:
- Put good in and get good out. Be a good person! Only add valuable content to AI models that is (to the best of your knowledge) as accurate and unbiased as possible.
- Do your own research! Ensure the pieces you’re repurposing are free of bias, discrimination, sensitive content that may be harmful to others, or potentially copyrighted content that you do not have rights to repurpose.
It sounds like a big task, but if you’re willing to give AI a chance and generally have a decent understanding of right from wrong, there’s no reason why everyone shouldn’t be able to use AI language models ethically and conscientiously.
Lillian Minix is the marketing communications manager at Timmons Group. Contact her at lillian.minix@timmons.com.