AI Advice For Everyone
I completed the Deeplearning.AI course "AI For Everyone."
The course instructor, Andrew Ng, has shipped a lot of powerful AI products in his career. He has also led teams that are pushing the bounds of Deep Learning methods. Having a course delivered from someone who has commercial experience with emerging technology is something I hope becomes a standard. Nothing against pure academics, but I gain stronger wisdom from people who have attempted to implement.
Those who implement get frequently humbled, so they can paint a humbling picture of what is possible and what is not so reasonable.
If there is one takeaway from Andrew's course it is this:
Extremes are rarely useful, so it helps not to let yourself get lost in the hype or the scare of AI. The truth is that machine learning and deep learning will evolve your role and industry. Before having a strong opinion, shouldn't you or your team start tinkering so that you are better prepared to change? That seems like a worthwhile objective.
This course was more potent than the more technical courses on machine learning I have attended. The title of the course says it all. So many people can easily assume what AI is and where it is going, yet everyone could benefit from a refreshing update on the status of the technology and how we can assemble successful AI projects.
I updated my mental models around leading Machine Learning prototypes after taking the course. Below are some of my actionable notes taken from some of the student exercises. I hope you find them valuable and I also hope you make time for taking the AI For Everyone course:
What are highly repetitive workflows in your business? Of those workflows, which input > output relationships have the shortest duration? The answers to these questions are a great start for prototyping an AI project.
Be careful not to promote your AI transformation until after you have gained experience and competency through building and teaching.
There are many abstractions and toolkits available for you to begin projects with today. It is a good time to build upon the great work of a growing community. Here are popular tools to get started with:
Don't expect 100% accuracy with your machine learning capabilities when you get started. By principle you need time to ensure tweaking, auditing, and data cleansing is. There are many determining factors that you will not have perfect. The good news is that the Machine Learning project will help you spot what needs to be fixed and researched further.
You could have a short team brainstorm right now by considering this context and asking the following questions:
Here is a playbook for a company wanting to leverage AI. I believe a lot of people make mistakes by working on strategy with only opinion rather than getting their hands dirty before developing a crisp and realistic strategy.
Overall, you can have greater success if you have the following approaches: