All the talk about artificial intelligence and machine learning may make it seem these methods are easy to use. But good artificial intelligence(AI) is based on building up knowledge over time by practicing with data to learn what people do, and how to respond. The same is true for people – learning takes place over time and with a lot of practice.

Data scientists start out by working with businesses to identify problems that computers can learn to solve.

Critical factors in planning for machine learning and AI include:

  • Keeping track of what data is available at the time a decision needs to be made
  • Separating data into what it looked like before a decision was made and afterward
  • Not including the result of a decision (which is also stored in a database) as part of what is known before the decision was made
  • Transforming data into knowledge. For example, taking a birthdate and turning it into an age. Or taking a list of purchases and turning it into a lifetime customer profile.

Data scientists determine which algorithms are good starting points for machine learning. They work to gather data that gives a more complete picture of the real world so the machine learning can be realistic.

To implement a complete AI system, business people, data scientists, and computer engineers team up to solve more complex and subtle problems, taking advantage of systems that are continually updated.


People formulated the goals. People developed the processes. People gathered the data. People gave the computers data to learn from. And now, people are benefiting from the intelligence that has been built.

Now, the first analytics project that a company undertakes can be difficult because data needs to be used in a new way. Groups of people gather the data, but these people may not be used to working together. And the data probably needs to be modified from how it was originally stored. It’s tempting to use the first model so the project team doesn’t have to recreate the data input process.


But this human (and business) process friction is easier to smooth out if the repeatable data gathering and machine learning processes are resolved sooner.

And machines are hungry learners, so companies  are more likely to succeed if they build processes that let the machines keep up with the world.

An artificial intelligence system is not made from magic, nor is it created in an instant. It is the result of a gradual build-up of knowledge, people, and business processes, geared to solve a specific set of problems.

Last updated: 8/6/2024
Image source: Pexels.com