Guidelines, more than "Policies"
Generative AI is a fascinating, fast-evolving area that's been on every headline recently. We've toyed with many tools and experienced, in one way or another, that feeling of witnessing magic. At the same time, there are several concerns about the use of AI in our industry, ranging from copyright infringement and plagiarism, to simulation and data management, among many others.
Ethics is ultimately, about boundaries, the lines we draw to make sure we operate within the limits of what we deem acceptable. And we work in an industry where people put their money where our mouths are, so what we say matters, and whether or not we can be trusted matters too. We deal with trust, and that is mainly built with honesty.
These guidelines are an open conversation that aims to provide clarity and foster trust in our industry.
This document is a work in progress. You're looking at v0.1, July 2023.
The tool or the machine
The craftsman himself can always, if allowed to, draw the delicate distinction between the machine and the tool. The carpet loom is a tool, a contrivance for holding warp threads at a stretch for the pile to be woven round them by the craftsmen’s fingers; but the power loom is a machine, and its significance as a destroyer of culture lies in the fact that it does the essentially human part of the work. Friedrich Schumacher - Small Is Beautiful, Chapter 1
These times are pushing us to face the age-old challenge of making ourselves irreplaceable. We do this by knowing more about our craft, not less, by understanding the real changes in our work, and by valuing talent and expertise. Our craft is a human thing. What we do is about people. Our job will always be the human part of the job.
We use tools, that we create, to create further, and to make things beautiful and useful. But those things reflect us and how we behave, what we do. And what we do is experiment, learn, adapt, go back to what others did before, and draw from experience. Connect.
In your face, Skynet.
A proposed approach:
Set rules.
Be open about them.
Question them.
Iterate.
While we figure out what these changes mean and wait for those in power to catch up with this century's zeitgeist, we can bear our responsibility, define our criteria, set our limits and make them crystal clear for customers, partners, and providers.
This "AI Code of Conduct" can take many forms. Here's what we're doing at One Totem:
-Asking questions to figure out the "why" behind what we are doing.
When using these tools, ask yourself:
- Are you automating tedious tasks? Or replacing someone?
- Is the result better than doing it the good old-fashioned way?
- Who are you helping? Clients, staff, yourself, the world?
- Do you have the resources (time, budget) to keep people in charge of delivering creativity?
- Are you optimizing something that doesn't need it?.
If you're incorporating AI into your daily processes for the sake of technology alone, please reconsider. Tech is only a means and never an end.
If it's to achieve more with less time and effort, think carefully about who that benefits, and be transparent about it. The race to obsessive efficiency does not generally benefit creativity, exploration or innovation.
If it will impact other people's lives negatively, consider if there's any other way.
If it's because "Everybody's doing it so why can't we" remember that following trends is a personal choice, and there's no real value in being an early adopter of anything.
If it's because the resulting work really improves all the other very human stuff you do, make it work (and be transparent about it: you didn't cook the chicken, you ordered delivery).
-Being transparent about what we are doing.
Tell your customers when something was built with AI. Articles, images, videos, descriptions, customer support, order processing, live chat, product descriptions, search suggestions, etc. Be open about the tools you use, and about why, and how you use them.
What we use it for
Double-checking stuff
Creative writing remains exclusively in the realm of our team's imperfect and wonderful organic brains.
However, most of our team members work in a foreign language on a daily basis. Some of us in more than one! For general communication and operations, ChatGPT, Writer, and similar tools can be useful for checking basic grammar, spelling and style, and planning layout or structure. We still use Hemingway and Grammarly for this, too.
Automating non-value tedious tasks
We still write our emails. We use AI tools to summarize conversations in order to prevent miscommunications. We are using it to suggest Excel formulas, write legal disclaimers and unorthodox lorem ipsums or do basic general research. And for fun too. We use it to choose random words for our weekly remote Pictionary team games.
Coding
Though we don't use it as much as we could, it has proven useful to find bugs and suggest workarounds or solutions. We're still scratching the surface on this. Truth is, it's becoming a substitute for StackExchange. Architecturing or designing solutions will stay with Team Human.
Illustrating
This one treads the fine line of the ethical dilemmas we discussed earlier. We like to experiment, and we don't have the budget, time or talent to paint cryptic 60s sci-fi custom images for our blog. The results are usually far from what we initially envision, but still cool and distinctive enough. We use AI generated images to illustrate our blog post and newsletters.
Debating
"Hey, Hal" (we like naming it), "I'm confident I can prove that Rock Lobster is the best song ever written. Please counterargue!". It's useful for finding gaps in your own narrative, or questioning preconceptions. This is particularly useful at 2am when you can't find anyone ready to confront you or bounce ideas off.
Bridging ideas
You know when you have this intuition that there are certain patterns between seemingly separate things, but you don't know enough to find them? Future Ultron does. It can join the dots faster. Find a few edges and ask the bot to find the common ground, and justify the argument.
Running stats
We're looking into using AutoGPT to run data and analytics check-ins on our properties, building custom reports with tailored data period comparisons and automated formatting. Still in the works, we'll share when we get there.
We will not use it for:
Faking insight
Any and every opinion across every brand is ours, backed up by our personal and group ideas, experience, knowledge and critical thinking.
We do not use it for brainstorming. We don't believe in using software to find us ideas. Our brains, our storms.
Faking work
Our team is instructed to avoid using these tools in any way that's not explicit to the customer. As a client of any One Totem brand you should be notified if our team decides, for any reason (pragmatism, experiments, budget, deadlines, etc), that we need to leverage these tools for faster or better results.
Faking personal voice
While we cannot and would never control what our team members say on work-related social networks (like LinkedIn, Reddit, Quora, Twitter, Mastodon and similar) we do not endorse AI generated posts on personal accounts.
We strongly believe in saying things with our own voice.
We still have no clue.
This remains an uncharted territory with more questions than answers. Experts don't even know how to evaluate LLMs performance (Azhar & Warren: Part 1, Part 2 & Part 3). We're taking adoption slowly, but fully aware that the technology is here to stay.
We set aside a budget of hours for our team to play and break things. We have an internal forum where our team members share their findings and discuss pros and cons. We're asking everyone for input to understand the role these tools may have in our daily work. It will continue to be a team effort, and a team discussion.
All in all, we're curious, but not hyped. The general feeling is we expected Jarvis but got autocomplete on its first day on the job.
It's a brave new world, nonetheless. (That has such people in it).
The map is not the territory
“. . . In that Empire, the Art of Cartography attained such Perfection that the map of a single Province occupied the entirety of a City, and the map of the Empire, the entirety of a Province. In time, those Unconscionable Maps no longer satisfied, and the Cartographers Guilds struck a Map of the Empire whose size was that of the Empire, and which coincided point for point with it. The following Generations, who were not so fond of the Study of Cartography as their Forebears had been, saw that that vast map was Useless, and not without some Pitilessness was it, that they delivered it up to the Inclemencies of Sun and Winters. In the Deserts of the West, still today, there are Tattered Ruins of that Map, inhabited by Animals and Beggars; in all the Land there is no other Relic of the Disciplines of Geography.”
Borges, “On exactitude in science”. 1946.
AI has been around for quite some time. The new thing is we’ve fed the LLMs massive amounts of data (our data, btw), and it worked well beyond expectations. The end product is an astonishing simulation of, well, us.
But simulations age badly, or end up becoming a parody. Creation requires experience, it comes from life. Its outcome is the interruption of repetition, almost a miracle.
If you have something to say, don’t let it be ruined by AI’s beautiful and pervasively useful imitation of creativity. Try it, test it, use it, but draw your own lines and proceed with caution. Own your thoughts, critiques, mistakes, and discoveries. Celebrate and honor the unique, irreplaceable, persistent relevance of your talent. The competitive edge is quality, not volume. The real prevails.
Send comments, suggestions, ideas, criticism or feedback at hello@onetotem.com
Made with ❤️ across many timezones
© Red Forest S.L. All rights reserved