AI and The Idea Center
We value transparency, and our clients deserve it. So, here it is: we use AI every day at The Idea Center to do more for you. This means streamlining campaigns, advancing SEO, and generating a high volume of ideas to move your mission forward.
AI as a human-driven tool
We view AI as just that: a tool. And, like it’d be unwise to leave a power drill unattended, generative AI models need a person in control. As Founder and President Barry Martin puts it, AI is like a typical software on steroids – and all this potential power makes human oversight all the more necessary. AI output requires supervision, prompting, and a human touch. Without our fingerprints all over them, responses are formulaic and predictable.
We’re also treating AI as a thought partner, a springboard, and the devil on our shoulder. Mark Deutsch, VP of Sales, argues it shouldn’t be an “end designer,” and Barry agrees, defining AI use as a back-and-forth conversation during which we massage, fact-check, refine, and fact-check again. When we partner with AI, it’s a second set of eyes to identify blind spots, record key details, and generate initial ideas. And though the constant stream of affirmation from an AI is certainly validating, it’s limiting, too. When every idea is a great idea, there’s little incentive to dig deeper.
Using AI intentionally
To keep users engaged, AI models tend toward agreement. It’s like asking your grandmother if you’re good at singing – of course she’ll say yes. But unlike your grandma, AI is capable of critiquing you. To get constructive criticism from your model, take a page from Mark’s playbook.
Mark has a specific folder containing his bio, unique writing style, and instructions to critique rather than constantly support. Mark encourages his AI to act as a contrarian by challenging assumptions and pushing back, and before you know it, your doting grandmother seems to have a few issues with your singing techniques.
Now that we’ve established how to mold your AI’s responses, let’s discuss the art of prompting them. Mark points out that prompts are evolving – they’re no longer limited by a keyboard. Instead, methods like voice dictation allow for natural, conversational inputs, resulting in free-flowing dialogue and rendering prompt engineering obsolescent.
In addition to how you interact with your AI, it’s imperative to choose the best model for each task. We’re fond of Google Gemini for concept ideation and generation, as we’ve found Google prefers its own content. And, because Google (as we all do) likes to hear itself talk, Gemini’s content is more likely to end up on the first page of search results.
In a tale as old as time, Gemini and ChatGPT are mortal enemies — and Google knows if you’ve consulted its nemesis. As Barry warns, “one thing Google will not let you do is cheat the system.” And, we’re not trying to. Gemini’s also great for visual concepting, especially on the fly. When a shadow of an idea takes shape during client Zoom calls, Barry prompts Gemini to generate a comp or two. Instead of taking a week to create detailed mock-ups that may be vetoed, he offers an immediate glimpse into a concept.
Anthropic’s Claude, on the other hand, lends itself to technical and SEO tasks. The counterpart to Gemini’s outward-facing assistance, Claude’s great for back-end utilizations like meta descriptions and Search Engine Optimization. Of course, all output is double- and triple-checked by our talented team, but allocating monotonous, repetitive work to AI models saves us time and, in turn, our clients money.
What AI can’t replace
Like that power drill, an AI model is only as effective as the hands controlling it. It requires human bookends: direction up front and refinement at the end, which often sparks another round of guidance.
AI is powerful, but it’s no substitute for human guidance and fine-tuning. It also can’t replace authentic content. Search engines like Google “want real people, real locations, real jobs, [and] real content,” Barry reminded us.
These models also can’t replace a human voice. As much as we love the em dash and the word “tapestry,” we recognize they serve as conspicuous indicators of generated content. AI writing continues to improve, but paradoxically, it’s getting easier and easier to spot. This blog, as we hope you’ve realized, is entirely human-written. We did, however, use AI to transcribe interviews, pull out key points, and organize them. So, AI helped compose a blog about AI. How meta is that?
AI and the future
The Idea Center is far from oblivious to the environmental impacts of AI. With these in mind, Barry outlined his hopes for the future – like utilizing geothermal power to run data centers. Insufficient energy sources are one of many growing pains associated with AI, and these road bumps, Barry notes, are inextricable from any technological innovation.
Feeling unsure about AI is natural; not everyone’s ready to dive in headfirst. But like Mark’s custom model, Barry offers some pushback. “If you don’t embrace AI, you’re going to lose your job,” he cautions. “It’s like people not wanting to embrace cars,” he says. “Sorry, we’re going to drive a car instead of taking the horse.” And, like the invention of the car improved the day-to-day for all but the most devoted equestrians, we believe AI has the potential to make lives better.
To experience how partnering with AI makes us a better partner for our clients, contact us today!