What to Make of AI Today: A Mighty Roundtable Discussion

How are you currently using generative AI in your work?

Andrew Buck: I use two AI-powered tools to help with my content work. The first is ChatGPT, which I use about twice per week to help me generate ideas. The second is Writer, which is incorporated into my favorite software, Google Docs, and which I use to perform a sort of advanced copy edit.

Stephen Tidmore: We’re currently using some AI coding tools such as GitHub Copilot and a new AI Assistant in PhpStorm. These help with some more routine coding tasks and code completion, but they can also help analyze code, make suggestions, and even document changes that were made.

Adam Wakefield: We built a lead scoring system to analyze our incoming prospects. That’s been super useful to help make sure we’re engaging the right audience. We also use HubSpot’s automated workflows for lead management, client communication, marketing, and content distribution in nurture sequences, and event registration. The successes are largely operational.

ChatGPT: Organizations are leveraging generative AI in various ways to enhance their strategies and productivity. Here are some common ways: content creation, chatbots for customer support, email marketing, social media engagement, market research, interactive advertising, content ideation, personalized recommendations, and training sales teams.

What are the key impacts and challenges you’ve seen in AI?

Jamie Pratt: It hasn’t made an enormous impact in my area of expertise, which is analytics. I saw a lot of marketers try to use ChatGPT and other AI tools to do keyword research, create content briefs, or generate SEO-optimized metadata, but I’ve found those results lacking. It’s a way for digital strategists to do the bare minimum, which has never cut it in marketing. The challenge for good marketing professionals is to distinguish the quality of their work from that bare minimum AI output.

Adam Wakefield: I can see it having a big impact on higher education. Middlebury College uses a Google form to invite students and alumni to describe their university. Then they feed those results into Midjourney as prompts to create new image content. This has a big impact on a college that doesn’t have enough money to hire a graphic designer. It’s also a great way to keep students engaged and have their ideas be, quite literally, seen. The big challenge‌ is the time required to get it right. It’s not a plug-and-play tool.

Dan Puzzo: If these AI features live up to their potential, it could save everyone a lot of time and money. Rather than skilled web developers being burdened with monotonous tasks in their work, they can shift those responsibilities to AI and concentrate on the more complex and crucial aspects of their job. I think the biggest challenge is overcoming user skepticism. Some people are going to need assurances about AI’s shortcomings to feel comfortable using it.

Stephen Tidmore: The coding tools help a great deal with the speed of coding, generating ideas for how best to handle something in the code, and analyzing code. Our dev team works more efficiently. But AI is going to get things wrong. It requires us to continue doing our jobs and not rely on AI to do our work for us, but rather to assist us in doing our best work.

Chat GPT: Using generative AI can have significant impacts on various industries and applications. Some key impacts include automation of tasks, enhanced creativity, improved productivity, personalization, and innovative applications. Key challenges include ethical considerations, bias and fairness, lack of understanding, data privacy, over-reliance on AI, regulatory compliance, resource intensiveness, continual learning and adaptation.

Let’s talk about the ethics of AI in marketing.

Dan Puzzo: It remains to be seen exactly what standard the business world will land on. Will we treat AI as a “subcontractor” when we bid on projects? Or will we not disclose it at all in the same way we don’t feel the need to tell clients we use Adobe InDesign, for example? My stance is that until the dust settles, disclosing the use of AI to clients is essential, both morally and legally.

Andrew Buck: Is there an immoral way to use artificial intelligence? Of course. You could use it to purposefully do harm. But is it immoral to use AI to create things such as ads, website copy, social media content, etc.? I’d argue “yes, if.” Yes, it’s unethical if you claim—or imply through obfuscation—that you don’t use AI. It’d be unethical for us to use AI, and AI only, to make the things we make for our clients.

Stephen Tidmore: There are still considerations for what the AI tools have been trained on. Was it trained on code repositories that weren’t intended to be used in this way? We also have to consider whether AI will generate the same ideas over and over again instead of coming up with new, creative solutions for the same challenges.

Jamie Pratt: The ethical pitfalls are staggering. You have organizations looking to cut their marketing teams, so job security is an issue. To echo Stephen, the training datasets for some large language models were scraped without the owners’ permission, which might be plagiarism on the greatest scale imaginable. Some AI tools‌ take user inputs and incorporate them into their Large Language Models, which can possibly break non-disclosure agreements or inadvertently leak sensitive intellectual property. The list goes on. It’s crucial that companies remain mindful and tread lightly with the technology.

ChatGPT: The use of tools like ChatGPT raises several ethical considerations that organizations should carefully address. Here are some primary ethical considerations: Bias and fairness, privacy, transparency and explainability, misuse and manipulation, informed consent, and impact on employment.

Where do you see AI technology going in the future? How should we prepare?

Jamie Pratt: It’s hard for me to say where it’s going. I think as professionals, it’s important that we carve out a niche where we can make the case that our services, our analyses, our writing, is better than what a client can get from an AI-assisted alternative.

Stephen Tidmore: I think it’ll become a competitor for some of the work we do, so we need to figure out how to work with AI in order to solve our client’s problems more effectively than AI can on its own. But I still think we should continue to explore and test AI tools in all areas of our agency in order to not be caught by surprise.

Adam Wakefield: I think the data analysis capabilities will become close to perfect within the next five years. That’ll totally eliminate humans from ‌tracking and reporting altogether. For people like me, our job will be to interpret the output and make strategic decisions, which is the far more rewarding part.

Andrew Buck: Who knows? Humans are terrible at predicting. But I suspect there’ll soon come a time when organizations with big enough budgets will spend real money on AI tools that do things such as: create marketing messages, target ads in a hyperspecific way, and then buy and place ads across the Internet. But these ads will be bad. My guess is that AI-made marketing content will work only on those who are “very online,” very bored, or very undiscerning.

Dan Puzzo: The advent of AI in business is like full-size tablets in cars. Automakers put tablet-sized screens on every dash and removed physical knobs and buttons. It turns out people strongly prefer the knobs and buttons, so now they’re coming back. Similarly, I think we’ll see a huge shift in the agency world towards relying more heavily on AI, only for leadership to rein it in a bit once they realize the quality of work in some areas will suffer.

ChatGPT: Predicting the precise advancements in AI technology over the next five years is challenging due to the rapid pace of innovation. However, based on current trends and ongoing research, here are some predictions for what AI technology may be able to achieve: enhanced natural language understanding, continued progress in deep learning, advancements in computer vision, explainable AI, AI for personalized healthcare, AI in edge computing, AI-driven personal assistants, new AI governance and regulation.

In Closing

Generative AI is still very much the mysterious waters on a 16th-century map—here be monsters. It’s impossible to say definitively whether it’s a gold rush or a cliff. But we know two things for sure. First, it should be reckoned with and explored as a tool. Second, the human costs and benefits should be weighed carefully, frequently, and transparently.

Here at Mighty Citizen, we’re developing an agency-wide policy for how we leverage AI tools. Until we determine the broader usage parameters that align with our values, we’re committed to the limited use of AI tools on simple, repetitive tasks and initial idea generation. To chat with a real and mighty human about whether or not AI tools are right for your organization, reach out to our team—we’d love to discuss.