ACEC recently hosted the second webinar in its four-part series on Artificial Intelligence (AI). The session, titled AI for AEC: Content Development and Application Integration, focused on how transformative technology is being (and can be) used for business development and marketing functions within AEC firms.
The session featured two speakers, Substance151 Creative Director Ida Cheinman and Sobo Chief Marketing Officer Frank Lazaro. Cheinman led off the webinar by noting that AI has been used in marketing – and in life – for years. “Think Google Maps, writing assistants like Grammarly, Chatbot, Siri and Alexa. It’s ALL AI,” she said. The breakthroughs late last year in generative AI were a watershed moment in the technology, shifting the ability to use them well from a nice to have to an imperative. “Everyone is using these tools now or will be in a short time. You must learn them to stay competitive,” she said.
The challenge with AI is the speed with which it is evolving. To demonstrate, Cheinman pointed to how long it took other high-profile platforms like Instagram and Airbnb to reach one million users. Instagram took two and a half months; Airbnb – two and a half YEARS. For ChatGPT, the one million user mark was hit in five days. What took other platforms months – years – to accomplish, ChatGPT accomplished in less than a week.
That kind of success breeds imitators, and they are everywhere. “The biggest challenge in marketing has always been differentiating new shiny things and substance,” Cheinman said. In choosing what platforms are right for your firm and its needs, she advised not to begin with tools and apps. Begin instead with what you are trying to accomplish. You might find the answers are already in front of you.
“Most existing platforms used by marketers have already integrated AI into existing features, or they will be soon,” Cheinman said. “Audit your existing platforms to see what you already have.”
The key to successful integration of AI is training the user. AI is only as good as the person directing it. Cheinman concluded, “I find that if you understand the basics, you can get really good at this really fast. You need to provide specific instructions on what you are looking for.”
Lazaro echoed that assertion throughout the second half of the webinar, stressing that whatever the AI platform, a human is needed throughout the process. “It’s called ChatGPT for a reason,” he said. “You really do need to have a conversation in order to get the content that you need out of it.”
Without the right prompts, he said, a “garbage in, garbage out” scenario is not only possible, but likely. “Poor instruction will lead to poor results,” he said, giving an analogy: “Think of generative AI as if it was an intern on your team. If you told them to go write some LinkedIn articles and gave no further instruction, what kind of results do you think you would get back? The results would be all over the place.” But, Lazaro continued, if you gave that same person a more prescriptive instruction, the results would be much closer to what you intended. “How you give instruction to AI is going to dictate the quality of the result.”
Lazaro went on to share the platforms that he uses in his work and gave a series of pointers on how to optimize AI prompts to get the best results. “Training AI is the most important thing you need to do to get really good content.”