Skip to Main Content
Home » Blogs » Your Bot is Only as Smart as its Teacher


Your Bot is Only as Smart as its Teacher


Last week I had the opportunity to attend the Intelligent Automation Exchange in Miami.  It was an interesting event with speakers from Google, BD, Mars, Uber, all whom you would expect to have much to say about the topic of Artificial Intelligence (AI) and Machine Learning (ML).

The day started with Cassie Kozyrkov, Chief Decision Scientist for Google. She was quick to point out two very basic facts about AI and Machine Learning that are often misunderstood or overlooked.

1. You cannot remove humans from AI OR ML. 

Many feel that the onset of automation, AI, and ML will displace many humans.  However, people really are at the center of those initiatives.  You must remember that the objectives, datasets, tests, and constraints of the systems are all set by humans.  Technology is only an echo of the wishes of the one that built it.  It is important to dream but dream responsibly. 

Which brings us to her second point:

2. AI & ML are merely students.  If they fail, it is not the fault of the machine but the teacher.

The data we provide must be unbiased.  To achieve that, we need to ensure that our data comes from a diverse set of sources.  If the student learns from a single book from a single author, all the views and understanding of the student will be shaped, for better or for worst, by that author’s views.  The data you present should come from and be validated by multiple sources. 

All of this however is jumping “waaaay” ahead.  It is important to note before you get to this point you must first identify what you want to automate.  Before you can do that, you have to know what you want the automation to accomplish.  To do that, you need to understand how the process is supposed to work.  The same rule Cassie applied to your data is also valid of your process.  If you only capture your process from a single point of view, it is not likely to fully represent what is truly happening or how things really work.  You must engage with individuals at all levels of the process to ensure your data is valid and unbiased.

The ancient Chinese philosopher Laozi stated that “the journey of a thousand miles begins with a single step.”  It should be noted that the first step should probably also be in the direction you wish to go.  Automation is a journey that can at times feel like it is 1000 miles in length.  However, if you take the first steps and ensure you understand, document, and validate the process, you will be on the right path and finish with a lot less detours. 

Another misconception is that RPA, Automation, AI, and ML will solve process problems and generate revenue.  I will discuss these in more detail in my next post.

Related Posts

ERP Giant SAP Puts Signavio in their Pocket. What Does this Mean for You?

  • clock January 28, 2021
  • author-icon Ed Maddock

For the past 30 years, iGrafx has championed how understanding business processes is central to a company’s success.  Documenting and capturing these processes while ensuring...

Read More

Perspective 6: More About Data (Part 2)

  • clock December 8, 2020
  • author-icon Paula Bell

In my last blog post we started discussing the topic of data. We discussed how to identify data, and one of the tools (Business Glossary)...

Read More

Perspective 6: All About the Data (Part 1)

  • clock November 20, 2020
  • author-icon Paula Bell

Now that we have wrapped up Perspective 5 around systems, we are ready to move forward with the final part of the Perspective Series, Data....

Read More

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.


Download our expert-led demonstration of the iGrafx Platform

Close Icon