AI & Machine Learning in a business

Posted by kdow on Feb 17, 2017 12:12:00 PM

The newest trend in modern companies, other than diversity in hiring, appears to be how best to leverage AI & machine learning. For some bigger companies, that means building some software on top of data to get better intelligence, insight & actions out of said data. Because pivot tables, SQL lookups and humans tend to make mistakes or lead to slow or inaccurate decision making.

As far as I can tell, there’s 4 main uses for AI or machine learning (I’m sort-of bunching both AI & machine learning into one box here):

  1. Optimisation: Using data to improve on some process. Simple example is finding the best route to work by avoiding traffic on Google Maps. A more complex example is using CRM data & past results to find the best leads for a sales team.
  2. Identification: Having the ability to dig into data & find patterns to identify objects or elements that are useful for decision making. Simple example would be Apple Images identifying faces to group photos of people into folders. A more complex example would be to identify trends in how a cohort of users interacts with new product features.
  3. Find problems: Humans are pretty good at identifying anomalous information. However there’s a distinct upper limit to the amount of information a human can take in, so using software to look at significant amounts of data to find issues can be a huge benefit. A simple example would be finding a bad fit lead in a CRM. A more complex example would be AI to detect credit card fraud.
  4. Segments: There seems to be an endless stream of data, lists, tables and cohorts to use in companies these days. Even my wedding list last year was comprised of 5 tabs in Excel featuring various data sets ranging from “will definitely attend” to “should be invited but definitely won’t attend”. Creating cohorts/segments is a huge part of machine learning, especially if some software can automatically generate perfectly curated target lists. A simple example would be my wedding list, solving the problem of who to invite. A more complex example would be segmenting a marketing database in real-time based on user interactions to a website.

All four of these are incredibly powerful examples of machine learning or using AI for complex scenarios anywhere (business or otherwise). But another, infinitely more powerful use-case is when you mix/match these examples.

This is the future of everything. All four elements above are used in self-driving cars, autopilot functions in aircraft, sales+marketing systems, or even complicated Excel spreadsheets. My point being that the future is complex, data-driven & exciting. And we’re in the midst of that future!