AI disclosure
AI plays a big role in how The HELIOS Report gets produced. On this page, we want to share more details about that and how we think about AI as a tool.
Ilene’s specialization as a faculty member was in rhetoric and composition studies, and she spent 15 years directing composition programs, supervising writing instructors, and teaching college writing.
Robert has been a professional writer and editor for more than 30 years and has delivered more than 2 million words of material to paying clients. He has won several awards for his journalism, has been anthologized in college writing textbooks, and has taught academic writing as an adjunct at four universities.
Those of you who took composition classes may recognize those last two paragraphs as a particular emphatic example of ethos — an appeal based on our authority. We begin with that to establish that we both are highly aware of the value of writing, share a deep understanding of the human intellectual work that goes into impactful writing, and are alert to the risks of careless use of generative AI tools. Like many others, we have observed that generative AI is far from capable of producing useful writing from a single prompt.
Nevertheless, we also believe AI provides an opportunity to extend and amplify the skills of experienced practitioners in a range of domains when it is used thoughtfully and with an understanding of the limits of the technology. So far we have found AI useful — particularly in producing The HELIOS Report — in these ways:
As a customized research assistant
AI supplements our regular practice of reading widely and critically to “keep up with the field.”
First, we use AI tools to help scan the environment and identify highly authoritative sources we may have overlooked that are relevant to ourselves, our clients, and the readers of The HELIOS Report.
Second, we use a highly customized LLM to summarize lengthy or complex materials.
Third, we use that LLM to suggest connections between disparate sources.
As a sounding board
We use LLMs as partners to explore alternative perspectives on topics we have identified as deserving inquiry. This process helps jumpstart our analysis and give us more time to deepen and refine arguments.
To assist in drafting
We use a highly customized LLM to aid in producing first drafts of The HELIOS Report.
“Aid in” in that last sentence is an important qualifier.
First, it refers to the methodical and iterative process that we find works best when it comes to creating original work with AI. However attractive it might be to give one prompt to generate one complete article, we find that approach simply doesn’t work, not even to generate a usable rough draft. We are much more likely to ask an LLM to generate one discrete piece of a planned article and to treat that as the first level of a scaffold that we continue to build on.
Second, “aid in ” refers to the so-called “human-in-the-loop” model. Yes, we are “checking the output.” More importantly, we have found that one iteration is far from sufficient to achieve the insights and clarity we expect from ourselves. (Robert says that model would better be characterized as “human-in-the-line.”)
In short, AI-assisted drafts of The HELIOS Report are just one stage of an extensive editorial process. We maintain editorial control over all content. We review every edition of The HELIOS Report to make sure it meets our standards for quality, and we take full responsibility for everything that appears under our names.
This AI disclosure statement was inspired by Michael Wagner’s statement on The Augmented Educator newsletter.