"The AI will do it."
This is a misconception Derek Ross, TrueMark’s Field CTO, hears again and again in meetings with executives. It’s the magic answer to almost every business challenge – even if it’s impossible or unreasonable to achieve. “Leaders need to understand AI because they can't direct strategy without it,” says Ross. With AI, in particular, it’s important to implement something that’s going to value in the business. “Leaders don't need to understand how to actually write an application, but they do need to understand the capabilities and limitations of AI,” he says.
Curiosity and continuous learning
Even though Ross is a seasoned cloud architect, he doesn’t have time to build workflows using AI, but that doesn’t mean he doesn’t use genAI (as well as other AI tools) outside of the office. “I use it to cook, to create my workouts, to build technical documentation,” he explains. “And that way, I can see where the limitations are and where I have to put in my own input.” In order to understand what new technologies are out there (and gain an entry level of understanding into how they work), making the time to try things out can help business leaders spot gaps, ask smarter questions and stay up to speed with what’s actually possible. “They don’t need to dive fully into the tech, but they do have to be playing and experimenting a little bit,” advises Ross. “You have to make time for research so you can have better and more informed meetings.”
For Ross, having a jack of all trades view of AI instead of being an expert in one domain is hugely beneficial – especially when it comes to setting policies around how AI is used within a business. “When a leader does know how AI works and they’ve had hands-on experience, they can say, very tactically, ‘you’re going to use ChatGPT for these three use cases’ or ‘we’re going to purchase GitHub Copilot for coding',” he continues. In Ross’ experience, the most forward-thinking leaders are the ones who are already creating custom-built applications in a service like Amazon Bedrock. “They know that they need AI in the space where they have PII or sensitive data,” he says. “They’re bringing that capability into a space that a lot of leaders are scared to do because they don't fully understand the landscape.”
Welcome to the age of AI
There is no easy button for AI. A decade ago, everyone was focused on DevSecOps and how, by being agile, that would solve everything. There were so many tools and technologies at the time but looking at DevSecOps today, there are two standout options for source control management: GitHub and GitLab. “We are at the beginning phases of AI,” says Ross. “There are thousands of tools out there and it's a very confusing space to be in.” Everybody's trying to get a piece of the AI pie and all the companies selling these solutions are doing everything they can, from a marketing perspective, to get businesses to buy in. “And this is why you have to have the knowledge in order to get past all of that,” adds Ross. “It’s complex, the dust hasn’t settled and every week there's a new architecture or tool or way of doing things. Leadership has to stay on top of this stuff.”
AI is everywhere… but is the real challenge technical or is it about leadership? “I’ve found that most leaders are going to OpenAI, negotiating a contract, going back to their team and saying ‘hey, I gave you guys ChatGPT, have a nice day' and then away,” says Ross. The correct way for leadership to implement AI is to work backwards. It starts with the problem you’re trying to solve and understanding the limitations of AI.
“One of the biggest gaps in business is the lack of understanding. They're just dumping everything into one AI model without actually understanding what underpins it or what it's good for,” he says.
Leaders, in particular, attach to one AI tool. Ross sees this happen a lot with ChatGPT and Anthropic’s Claude. The problem is that if a business needs to crunch large data sets, generative AI isn’t always the best fit. “People are chasing insights out of LLMs and GenAI, but that’s where machine learning and statistical analysis belongs,” says Ross. “Leadership is about knowing when generative AI isn't the right tool and knowing what scenario to use the right tool for.”
Share