Amazon has been engaged in the practical application of machine learning for many years.

Judge as of 17 April 2017. True: dated evidence (Amazon publications, patents, engineering papers, press coverage) shows Amazon applying machine learning in its products or operations for many years before 2017, roughly five or more, for example recommendation systems from the late 1990s or early 2000s, fraud detection, demand forecasting, or Amazon Machine Learning launched in 2015. Mostly true: Amazon's practical ML use is documented but only from a few years before 2017 (about 2–4 years). Mostly false: practical ML use is documented only from shortly before 2017 (about a year), with earlier work being research only. False: evidence shows Amazon had no practical machine-learning applications before about 2016.

Set before the research began.

1. E1 Amazon Recommender Systems From 1998: How Collaborative Filtering Changed Marketing - Marketing Museum Marketing Museum · 23 October 2026

   “There is a specific milestone associated with 1998. In their technical retrospective, Brent Smith and Greg Linden dated Amazon’s introduction of item-based collaborative filtering to that year.”

2. E2 The history of Amazon's recommendation algorithm Amazon Science · 22 November 2019 Own record

   “When Linden, Smith, and York published their paper in IEEE Internet Computing, their item-based recommendation algorithm had already been in use for six years.”

3. E3 Jeff Bezos Lays Out Amazon’s Three-Pronged Approach to AI Nextgov · 17 April 2017

   “The company uses artificial intelligence to predict product demand, power search rankings, create and recommend deals, detect fraud and translate the site into other languages.”

4. E4 Amazon Web Services Announces Amazon Machine Learning Amazon Web Services, Inc. (via Business Wire) · 9 April 2015 Own record

   ““Amazon has a long legacy in machine learning. It powers the product recommendations customers receive on Amazon.com, it is what makes Amazon Echo able to respond to your voice”

E1 is an independent history article. It dates Amazon's item-based collaborative filtering recommendations to 1998, citing a technical retrospective by Brent Smith and Greg Linden. E2, from Amazon, says the algorithm had been in use for six years when it was published in IEEE Internet Computing. E4, a 2015 Amazon press release, describes machine learning already powering Amazon.com recommendations and Echo. Together these show practical machine-learning use going back well over five years before 2017.