How do AEC industry businesses use artificial intelligence effectively and responsibly for writing?
The glitz and glam of artificial intelligence tools have hit the business world with a head-on tackle that would make even the best NFL linebacker jealous. The promise of saving money and time with automation has some flocking to the ChatGPTs, Grammarlys, and OtterAi notetakers of the market. But, if you’re like me, you may feel some hesitancy toward the “AI solves all” trend.
Perhaps this uncertainty stems from being cautious about in vogue things being pricey fads with low rates of return or from concern about the unknown. How do we know that generative AI-produced writing contains accurate information free of plagiarism and problematic material? Who is ultimately responsible for the texts that go into generative AI databases and come out of AI machines? Who owns and can profit from AI-produced material?
As someone who makes her living writing and teaching writing, these questions plague me, and I can’t help but wonder if we’re opening the door to I, Robot or Millennium Man becoming our realities – neither of which I’m excited to see come to fruition (although they are good movies).
As a bit of background, generative AI finds information in a database, puts it together, and produces a combination of output in line with what you request of the AI. For example, I could ask ChatGPT for a technical report on acid rain’s three most harmful effects. With one click, boom! There’s my report. But therein lies the problem: Is the technical report really mine? Is the information correct? Is it something that my readers will positively respond to?
Many public-facing generative AIs’ databases often consist of the entire internet. Everything that was, is, or has been posted on the internet is supposedly fair game. That’s a lot of material.
But what about those authors who don’t want their copyrighted texts subject to being plagiarized in my technical report? Too bad.
What about those who don’t want their information being used by someone they’ve never met for whose purposes they may not agree with? Sorry, you’re out of luck.
What about the information that’s biased, hateful, or just plain wrong? Well, you better hope that’s not what the AI creates for you.
Perhaps my bias is showing, but AI-produced texts can go horribly wrong.
This isn’t to say that AI should be avoided at all costs in writing. AI is here to stay. I won’t pretend to be a fuddy-duddy who refuses to get on the AI train and who yells at neighborhood kids to get off my lawn. Yet, AI cannot be without human intervention if we’re to create texts that matter, are effective, speak to our readers, and don’t get us into trouble.
Recently, in Thaler v. Perlmutter, a federal district court upheld the U.S. Copyright Office’s refusal to register a piece of AI-produced art because its origin wasn’t human enough. Essentially, a “natural human” must be the bedrock of a piece of art or writing for it to be considered for copyright. This legal precedent sets the tone for how America will view AI-produced texts. It shows that for something to be considered with ownership, humans must play an active role in its creation.
Will texts produced mainly (or even only) by AI show up? Yes. I’ve read those one-note and boring AI articles, and they made me question the companies I saw attached to such muck. To use AI and eschew the human touch is to discredit one’s company and oneself.
Businesses that wish to use AI in producing internal or external documents, in taking notes, or in data mining cannot expect to purchase the AI to ensure all will be well for their teams. Doing this will only throw team members to the robotic wolves.
Organizations that handle sensitive or private information and/or wish to safeguard their intellectual property should be cautious when using AI. Once you put information into the AI’s database to be spell-checked, organized, or whatever else, it’s there. Forever. Your mother warned you about putting things on the internet, remember?
So, what do we do? What’s the solution to using AI responsibly and effectively?
- Companies can anonymize or eliminate sensitive information when using AI.
- Businesses can go down the pricey path of creating their own generative AI that has access only to an internal database of information. Nothing but approved material goes into the AI and, theoretically, nothing gets loose on the Wild West of the internet.
- Companies must train their teams to use AI. This training includes crafting effective prompts, developing privacy and intellectual property issues, and using critical thinking and problem-solving skills to edit writing, ensure that writing meets the writer’s and the organization’s desired outcomes, and craft writing so that it speaks to the writer’s target audience. This training can then be translated into clear guidelines.
Businesses and writers who use AI without understanding how to apply critical thinking and problem-solving skills to writing will ultimately fail in their writing and AI endeavors. Bottom line: If a team’s writing is bad now, AI won’t fix the problem. The writing will stay poor and may get worse with the added complication of reliance on AI.
AI must be supervised and used in conjunction with human creation and critical thinking. We must view AI as a tool, akin to the typewriter or computer word processor. In short, AI cannot be the sole or primary creator, and it cannot replace sound human thinking and writing.
Elizabeth Preston, Ph.D., is an executive consultant for Hurley Write and the producer and co-host of The Writing Docs podcast. Connect with her on LinkedIn. For more information, contact info@hurleywrite.com.
