Roundtable Participants
- Jen Molnar, Director of Brand Content
- Brian Dean, Director of Digital Strategy
- Anita Holman, Executive Creative Director and Brand Strategist
- Chris Graver, Associate Creative Director
- Nick Booras, Director of Client Services
- Sandy Embrescia-Hridel, Director of Strategic Insights
Q: What does “AI” mean to you? What picture comes to mind?
Jen: It's not any singular tool like ChatGPT or Midjourney. There are a lot of AI features built into things we all use every day, like predictive text on your phone, navigation apps, or personalized recommendations from Netflix or Amazon. It's a resource—like electricity, in a way. It will just be built in to how things work.
Brian: AI means efficiency. Efficiency to quickly consume large amounts of data and quickly provide an output that would take a human weeks to complete. Once information is digested, AI then allows the end user the ability to automate tasks to drive further efficiency in their day-to-day lives.
Anita: On the one hand I love how AI helps serve me exactly what I want or makes my life easier and more efficient. On the other hand, I am concerned about how far it will go and who will eventually regulate. When thinking of a visual that embodies AI right now, I think an iceberg is the perfect analogy. You have this beautiful ice mountain above the waterline. However, below is murky, with little visibility to the size or mass. That's how I feel about AI, the great unknown. It's hard to tell how it will truly impact me personally or professionally.
Chris: Just like any tool, it’s only as good as the craftsman using it. The amount of time, knowledge and effort you put into it, will yield the different results. A hammer won’t work with a screw—knowing when and where to use AI is a skill in itself.
Q: Should agencies be allowed to use AI? Is it “cheating”? Why or why not?
Jen: I think Chris' response is a great start to answering this. It's one more tool at our disposal. It's also important to answer in the context of "What is AI?" Because it's already being built in to so many tools, it would be nearly impossible to not use it at some level. Should we be deploying fully AI-generated campaigns? No. Should we leverage various AI-enabled tools to explore other angles, ideas and insights, and to streamline perfunctory tasks so we can focus on more value-added activities? Yes, I think so. I don't think it's going to be a question of whether agencies use it or not, but how.
Nick: I don’t believe the use of AI by brand and marketing agencies is cheating; it’s strategic adaptation. It streamlines processes, allowing marketers to focus on creativity and strategy. AI also provides deeper customer insights, leading to better decisions. But agencies do need to use AI ethically, which means ensuring transparency and privacy. Rather than cheating, it’s akin to using a powerful tool to achieve better results.
Q: Is there a comparable sea change for agencies that you could point to in the past few decades? A technology that also really transformed the work? Or is this just in another league in terms of impact?
Jen: The internet. When I started in this industry, some companies were just getting email, there was no Google search, you needed an AOL keyword or the direct URL to get to someone's website and what was there was VERY basic. But before long, many printers were put out of work by the transition to digital. Brochures haven't gone away but a lot of the info you used to need to get on paper is now instantly and always available online. Some printers adjusted to also offer digital printing and incorporate other features, but it definitely did change the landscape.
Brian: In terms of AI's use within agencies, I think it is important to understand that AI is only a tool and not a solution for success. The human mind has the ability to strategically think about a problem and craft unique solutions to solving the problem. AI is only as powerful as the inputs it receives and the data that it has access too. It's not as simple as posing a complex problem to AI and getting a full strategic plan that fits. I personally compare it to other tools that have been developed to help agencies tackle marketing, such as CRM/email platforms, analytics, DMPs, etc. These are all tools that were developed to help with speed to market, but all of them will fail if you don't have a strong strategic approach to how and when to use them.
Q: Fair point. But with AI advancing exponentially in the coming years (according to some of its creators), what do you see as the risks/worst outcomes?
Brian: My biggest concern with AI is the ability to spread fake or incorrect information in a more powerful way through a multitude of content forms. I immediately think of deep fakes and the speed to which these things are being used. More concerning is the end-users inability to identify if the information they are receiving is real or fake.
Q: Conversely, what do you see as opportunities with AI? What are some surprising uses people may not yet understand?
Sandy: AI is being tested and learning to predict consumer responses and behaviors in many different industries. In marketing and advertising, AI tools are learning to more accurately predict performance of campaigns, images and videos with the expectation that the AI tool will be able to reveal how audiences feel about ads.
AI tools are also learning to act as synthetic respondents for surveys, focus groups, panels and the like. The results so far don’t make a good substitute for primary research with real participants and respondents. But AI tools can help build preliminary understanding of topics and issues so more advanced and relevant questions can be designed for humans. AI is better at answering “what and how” questions but still lacks depth when answering emotionally-oriented “why” questions, though that’s expected to change as it grows and gets more refined.
Q: Was a "synthetic respondent" on the wish list for people working in marketing research, or did it exist in another form?
Sandy: It's more nuanced. If we were a less experienced/more junior research and marketing team that did not understand our client and industry audiences so deeply, it may be more useful. It's bound to change as AI advances but today our baseline understanding before conducting primary research exceeds what we've been able to query from AI.
I think advancements in machine learning and AI, and the mass of data that modern AI can access quickly, has enabled an advanced theoretical definition of "synthetic respondents,” and a less developed functional application in practice, so far. There are probably some firms that do a way better job at creating and utilizing them, but I believe the practical creation and effective use of synthetic respondents is in its infancy.
Q: Nick Booras curious for you, pre-AI and post-AI, how you feel about those new opportunities and their impact on the worlds of marketing and building materials? Does one in particular stick out to you as an opportunity and/or have you seen it in action?
Nick: Automation and process optimization stand out. Automation can help lead to getting more out of business verticals like manufacturing. More time and throughput. Capacity models and needs for larger footprints become less, and seasonality in a business becomes less of an issue.
Process optimization and using something like visual AI helps recognize sub-optimal behavior. A close example would be at an auto body repair shop and an improper estimate being done by a human vs. AI. For a car to be repaired back to its original state, an accurate estimate needs to be done ... parts needs, labor time, paint, calibration, etc. If an estimate is wrong, in most cases the insurance company approved an original estimate and now, the body shop repairer has to go back and ask for more money, leading to longer repairs and more money towards a rental, resulting in an insurer having to raise premiums and ultimately put a policy holder in jeopardy.
A closer example would be estimates done to homes as the R&R segment grows. A lot of human power goes into these estimates and can require fuel, car, a human being to do the estimate and close the sale. All of this can go away eventually with AI and a customer could eventually self-service thru platforms and pick their product, contractor, etc., ultimately making it cheaper and more valuable for the consumer.
And I think building product manufacturers can see a ton of benefit by improving distribution as AI can deliver incredible disruption longer term in this area. Turning inventory in close to real time could lead to enhanced value, warranties and real time distribution. The patents being sought after now show what the future might look like down the road.
Q: AI brings with it the potential for virtually unlimited volume: untold levels of images, written content, etc. that exists merely because it’s possible to create, not because it’s good. How do you see that impacting marketing and branding specifically and different companies’ positions on the internet, their ability to draw people with meaningful content?
Jen: From a writing perspective, I think there's real potential for a large volume of very average content. The way these large language models (LLMs) work is that the algorithm is always trying to predict what comes next based on the examples it's been trained on. So it's going to produce writing that sounds like other writing. In order to create differentiated content, we'll still need people to continue to bring that innate creativity, cultural nuance and unique perspective that comes from the human experience.
Q: What’s already here in terms of AI technology that’s potentially being under-utilized or is under the radar?
Anita: One valuable characteristic of AI is the ability to consume a lot of information and quickly and easily sort through and disseminate insights or answers. Where I see this being leveraged is in training, customer service and field repair. Think about the amount of time it takes to train customer service reps. Even after hours of training, questions come in that require research or digging to find the answer. Now imagine if there were a data base of previous questions/answers, installation or product manuals or even design recommendations available in seconds to the rep. Producing answers or helpful suggestions on one call will be satisfying for the rep and the caller and probably improve the overall CX of the company. This training approach could be extended to sales teams or even customers, like distributors or retail associates, who are selling on our behalf.
A word on our AI Task Force
AI is already growing deep roots into everyday technologies we use and only growing. As an agency, we’ve organized a team among our staff members to address how our agency will use AI and to develop a deeper understanding of the technology’s implications for our clients. Check out principal and agency owner Bill Rossiter’s take on how best to approach staffing for these changes in his latest blog, “How to prevent humans from spoiling your AI decision making.”
