Five aspects to consider while managing AI projects

- Developing an AI project is not a technological challenge. It's far more than that. Let's now discuss the three aspects you must consider when developing an AI project. First, dream big, but start small. There is much discussion about AI's disruptive nature, especially generative AI. It has the power to change pretty much all dimensions of our lives. When envisioning an AI project, we must think this way. We must think big. When my partners and I envisioned the PMOtto, we envisioned a world where project managers could have on their side a virtual assistant that would change how they manage their projects with a massive amount of knowledge and combined project experience at their fingertips. The Standish Group says that, "Out of the almost 48 trillion invested in projects yearly, only 35% succeeded." The wasted resources and unrealized benefits of the other 65% are mind-blowing. Imagine the benefit we could deliver if PMOtto could reduce this by 1%. It would be hundreds of billions in results. But if we think this way, we may start with thousands of developers and massive investments, but, in reality, it cannot work this way. We must start small. We must test our assumptions. We must create MVPs that confirm our direction of travel. Think big, start small. The second is simplified governance. To move fast, we must have simplified an effective governance. We must ensure we do not create a complex layer of decision-making processes. They will only act as a break in your need to go fast. You must have a governance process that can act quickly and effectively, and this is not an easy task. It's very easy to go from 8 to 80, alternating a no-governance mode with a full cumbersome and bureaucratic governance at the other end. You must find balance. One tip that I can share with you is the principle of PRINCE2. PRINCE2 is an acronym for Projects in Control Environments, and it's a very popular delivery method for projects widely used in Europe. The principle is "management by exception." In this principle, you create thresholds and give the team the freedom to operate under those thresholds. If the threshold is exceeded, then the governance body comes into play. The third is algorithm selection. There are several types of algorithms and machine learning models. You must carefully select the one that is best suited for your project. It's not one-size-fits-all. They depend on the amount of data you have, the linearity of the information, the training time, et cetera. It's not just pick the most popular one. Moving to scalability, OpenAI ChatGPT got 1 million users in one week. And just to give you some perspective, Netflix took 3 1/2 years to do the same. Do you imagine how difficult it is to scale a solution at this pace? We must keep in mind that our solution may not follow a linear scale. Scalability must be a central part of your thinking to respond to this highly dynamic and volatile environment. Another critical aspect of developing AI is how future-proof your technology is. You must develop a platform that can quickly shift from one model to another. You must be prepared to switch if this is needed. And remember, the only constant in an AI project is change.