In This Issue
Vol. 25 Issue 38
The Power of Knowing Why
- Machine Learning – Powerful Predictors
- Neural Networks
- Why Is “Knowing Why” Important?
- A Second Example: The Power of Knowing Why
- The Next Trend in AI-Powered Industries
- In Summary
- Appendix: Sample Data from the Wisconsin Diagnostic Breast Cancer Dataset
- About Ty Carlson
Today, almost any company (IBM, GE, Google, Microsoft, Amazon) that says it’s using “AI” means it is actually using Neural Networks, a technology from long ago that found a new life not so long ago, for reasons that remain interesting and somewhat unclear.
The good news about neural networks is, if you know what you’re after, they can be trained to find it with high precision. Picture a bloodhound sniffing a dirty T-shirt and tracking the convict through the swamps until –
But the bad news is, as most everyone knows, this predictive ability is not backed by a simultaneous understanding of how that accuracy was achieved. This is often called the “black box” problem, and it represents a major failure for the technology, and for the companies and governments that use it.
A good example of this problem can be found in today’s GDPR laws in the EU, which require a business using AI (neural nets) to be able to tell a customer why that loan was declined by the system, that house not offered, that credit rating lowered.
It’s the law, and yet few (if any) firms can do it.
In this week’s issue, you will hear from Ty Carlson, CTO of Pattern Computer Inc., on the importance and benefits of moving beyond this mathematical roadblock. Scientific progress requires it, governments demand it, and you, as a user / citizen / customer, should, too. In the most important ways that society could imagine, this is where we need the Next AI to go next. – mra