It’s exciting times, and challenging times, for business. Everyone from the C suite on down is scrambling to figure out how to use brand new tools and ideas to their advantage.
For the rank and file, people below management level, the imperative is to justify their own work, by learning how AI applies to any given role (I cite Toby Lutke’s Shopify memo). Managers and leaders, on the other hand, have a slightly different goal – they have to figure out how to use AI to the benefit of the organization as a whole.
So how do you get confidence for, as a fortune cookie might say, these uncertain times?
Researching AI
One way to start is to learn about the technology in general, to start becoming knowledgeable on what the LLMs do, and why.
Just for example: I came across this list from Codemotion of common algorithm components and stochastic ideas using in AI/ML:
- Linear Regression
- Logistic Regression
- Decision Trees
- Random Forests
- Support Vector Machines (SVM)
- Naive Bayes
- K-Nearest Neighbors (KNN)
- Artificial Neural Networks (ANN)
- Convolutional Neural Networks (CNN)
- Recurrent Neural Networks (RNN)
- Generative Adversarial Networks (GAN)
- Reinforcement Learning Algorithms (e.g., Q-Learning, Deep Q-Learning)
- Clustering Algorithms (e.g., K-Means, DBSCAN)
- Dimensionality Reduction Algorithms (e.g., Principal Component Analysis, t-SNE)
Now, if you’re in a leadership role practically, it’s to your benefit to know and understand these terms. They represent a short survey of how we started using AI for things like restaurant recommendations, analysis of performance reviews, and decision support. In other words, if you can explain each of these types of machine learning mechanisms, you’re closer to the top of the pile when it comes to brainstorming on AI.
Tips from LLM Engines
What do MS Copilot and ChatGPT have to say about the issue? After all, you’re trying to understand them, in a way.
I got this by asking Copilot to enumerate some tips for better AI adoption:
- Define clear objectives and scope
- Adhere to a set of best practices to ensure future readiness
- Use AI to handle routine tasks and analytics
- Choose the right approach in preparing datasets and employing foundation models
- Produce powerful AI solutions with user-friendly interfaces and access to industry-standard APIs and SDKs.
(For sourcing, Copilot cited Codemotion and analyticsinsight, along with giants Microsoft and IBM, and our own work at Forbes.)
Then I asked ChatGPT and got this (I forgot to tell it not to be so wordy):
Start Small and Prioritize High-Impact Use Cases
- Don’t try to overhaul everything at once. Identify a specific problem where AI can drive measurable value (e.g., automating customer support, improving demand forecasting).
- Pilot a project in a controlled environment to prove ROI before scaling up.
Invest in Clean, Well-Organized Data
- AI systems are only as good as the data you feed them.
- Audit and prepare your data—make sure it’s accurate, complete, and accessible.
- Establish data governance policies to ensure ongoing quality.
Upskill Your Workforce
- Provide training so employees understand AI tools and how to use them effectively.
- Encourage cross-functional teams (e.g., business + IT + data science) to collaborate.
- Make AI adoption a shared journey, not something imposed top-down.
Choose the Right Partners and Technology Stack
- Evaluate vendors carefully—look for solutions that integrate with your existing systems.
- Consider scalability, security, and compliance (especially with customer data).
- Avoid lock-in by ensuring you retain ownership of your data and models.
Focus on Change Management and Clear Communication
- Communicate the why behind AI initiatives to build trust and reduce resistance.
- Address employee concerns about automation replacing jobs—highlight how AI can augment their work.
- Define success metrics and celebrate quick wins to build momentum.
After Human Review…
I was looking at how these pieces of advice overlap. You could say that “define clear objective and scope” from Copilot tracks to ChatGPT’s #5 tip on clear communication. As for “Choose the Right Partners and Technology Stack” (from ChatGPT) there’s Copilot’s exhortation to have the right tools on hand.
I’ll leave it up to you, the reader, to decide if these recommendations are overly generic or not.
More on AI Adoption
Then there’s this panel discussion from IIA, moderated by Paul Baier of GAI Insights.
“(You should have) practical applications of evolving technology, but also have dedicated focus on your own plan execution,” said panelist Venkat Vedam. “(You should have) use cases and business cases mapped out … this year, next year, but at the same time, we don’t want to lose on the skills gaps.”
The panel also discussed shadow IT, where employees may be using tools not explicitly endorsed by the organization.
“I feel like shadow IT is not a problem,” Vedam said. “It’s more of an opportunity … the reason shadow IT exists is because the employees are not getting the tools they need … it’s also manifesting in a slightly more structured fashion.”
He explained.
“There’s a bunch of engineers and developers who are servicing a small set of users that are not technically part of the technology organization,” he said. “The reason why the shadow IT works is (that) the people who are implementing those tools know the business problems well, and have the flexibility to adopt new technologies. (The goal is) to build an operating model around it … (and) to have a governance process to take what works in the shadow IT and make it real.”
“With everything changing so fast, I think it’s hard to (have shadow IT because) your organization doesn’t really want that so much anymore,” said panelist Joan LaRovere. “What is the problem you’re trying to solve? And … do we need to think about other vendors or internal builds? … you (should) know what you need in your tech stack to actually solve the problems your organization needs to solve, and you need that oversight.”
“I think what you’re trading off against is security,” added panelist Tomas Reimers. “And so if your employees are bringing in tools that have access to customer data or personal health information, that’s bad. If they’re using AI tools to make restaurant reservations for a meeting they have at noon, it probably doesn’t matter.”
The Spread of Information
Later, Reimers talked about observing tech processes and interactions to get a better bird’s eye view of what’s happening.
“One of my favorite graphs we have in the office is, whenever we go into an organization, we can actually map the social network of developers that talk to each other, one of the artifacts of working in development. And then you can see where it’s adopted. And it always looks like it starts at a node and it spreads out from there.”
LaRovere mentioned the value of broader collaboration, which is another point that resonates with me in terms of offering part of a road map.
“I think one of the best things … is bringing people together and sharing either what they’ve done, showcasing what they’ve done, testing different things, creating that, what we call a learning community,” she said.
Your Own Business Case
I’ll end with this: part of what I’ve learned over several decades of being around technology is that most new tools can either help or hinder a business (if you’ve read a good number of these blogs, you may have read this already) in terms of practical integration. There’s usually a learning curve. If you don’t prepare staff, you could be in for a lot of trouble. And then there’s fitting your applications to your business need, which is not a one-size-fits-all or cookie-cutter type of thing.
But maybe this set of tips, from people, the web, and LLMs, is a good start.
