10 Things You Need to Know about AI Right Now

Press enter to search
Close search
Open Menu

10 Things You Need to Know about AI Right Now

By Joe Skorupa - 05/04/2017

Artificial intelligence is at the peak of the hype cycle so you know what comes next, right? Some sort of trough or plateau maybe? Not this time. The reality is the famous hype cycle with its curvilinear shapes and text overlays is dead, killed by Moore’s law and the accelerating pace of change. Artificial intelligence (AI) is a perfect example of why.

I have been studying AI for more than two years and last week I took a crash course at the TCS Innovation Forum in New York and the Retail & Consumer Goods Analytics Summit (RCAS) in Chicago. Both conferences highlighted stunning revelations about the vast capabilities and rapid adoption of AI in retail and beyond.

I hesitantly use the term artificial intelligence, because it is often misunderstood and misapplied. However, AI works as a useful term when applied to powerful computing capabilities that use immense databases for prescriptive analytic functions coupled to autonomous actions.

For example, Amazon Alexa, Apple Siri and Google Home can listen to a voice command, understand the meaning of the language used, and then choose the best possible answer or course of action.

AI, used in this way, is not to be confused with voice command or natural language processing (NPL) alone. Instead, it is the ability to understand the meaning of the command by filtering out the noise of casual language. Then, after considering the meaning and weighing the possibilities, it responds by choosing the best possible answer or by taking the best possible action. One final thing, importantly, is that over time it tracks its success rate and prioritizes future responses based on the outcomes.

There is a lot of science and technology behind AI and you will hear such terms as machine learning, deep learning, autonomous intelligence, and augmented intelligences. In some cases, each of these terms can be applied to AI and even used interchangeably.

What really matters, however, is how AI is being applied in retail.

10 Things You Need to Know about AI

AI is being touted as the "next" big thing and it is, but "next" does not mean in the future. AI is functioning and influencing the world of retail right now. Here’s how.

  1. In addition to Alexa, Amazon uses AI for dozens of core retail processes, such as route planning for shipping and deliveries, optimized warehouse locations, automated warehouses for picking, packing and shipping, localized product mixes, recommendation engines, personalization engines, product testing, and more.
  2. Virtually all pricing decisions are made at Amazon by autonomous computing – AI. Insiders say that Amazon has shifted all former day-to-day pricing executives into other responsibilities and is in the process of doing the same with many merchandise managers.
  3. Walmart is not far behind Amazon and uses AI for supply chain planning, supply chain execution, logistics and replenishment processes.
  4. In addition to Walmart and Amazon, consumer goods companies are racing at full speed into AI and they are farther ahead of the curve than most retailers. Well publicized leaders in this area include Kimberly-Clark and P&G. Other companies in the retail landscape that are heavily involved with AI are FedEx, UPS, Facebook, Google and Uber.
  5. AI is hard for most retailers to master but not for the obvious reasons of advanced science and technology, which are indeed hurdles to overcome. AI is hard to master because it requires specialized data scientist, data engineers and AI scientists. These are among the most sought after experts in the world today. Not only do they require impressive salaries, but keeping them on board for any length of time is difficult for most retailers.
  6. Retailers don’t have enough data. Retailers have gone from possessing mountains of data to being data poor in terms of the amount of data needed for AI to accurately prescribe actions and become autonomous. Even the data retailers currently possess is not enriched enough or annotated in a way that links all the datapoints together with external data feeds from weather services, social media, search engine terms, sensor data (IoT), and reviews or ratings.
  7. AI requires a shift in the corporate mindset. We used to think of data as being well behaved, organized in rows and columns. Now it is unstructured and we need new tools and methods to manage it. Retailers used to be masters of rock-solid processes. Now they have to become masters of data,  which includes mastering of aggregation of new data sources, annotation and algorithmic synthesis between insight and action.
  8. AI is self-programmable. It is not only self-learning, but it can create its own algorithmic-driven bots without the help of programmers and these bots can autonomously perform complex tasks without instructions.
  9. Humans are not as good as computers at comprehending large volumes of data and creating data-based forecasts or demand-based plans. Initial AI use cases have shown that AI can significantly reduce time and increase accuracy for such key roles in retail as merchandise planning, assortment planning, allocation planning, demand forecasting, category management, pricing, campaign management, and promotion management.
  10.  AI is the third phase of the advanced analytics revolution. Analytics 1.0 is centered on reporting and descriptive analytics. Analytics 2.0 focuses on big data, data scientists, unstructured data and predictive analytics. Analytics 3.0 focuses on machine learning, automated decisions and prescriptive analytics.

These 10 things sum up the current understanding by the experts who appeared at the TCS Innovation Forum and RCAS about the impact of AI in retail, but it is only the start of a very short and rapid journey.

“By far the greatest danger of Artificial Intelligence is that people conclude too early that they understand it.” – Eliezer Yudkowsky, Silicon Valley AI autodidact

More Blog Posts In This Series

Searching for Retail Truth

Stores Are Dead, Stores Are Not Dead, Get a Grip!

Broad generalizations about the state of the retail industry absolutely positively fudge the truth.

Kidbox subscription box retailing

Kidbox: Disruption Through Machine Learning

What makes Kidbox interesting in the hyper-competitive retail environment is that it is disruptive in a disruptive category – subscription-box retailing. As with all areas of retailing today there is a proliferation of sub-box startups.

Which Killer App Survived the Retail Shark Tank?

Inspiring ideas that produce “aha” moments are out there but difficult to find, which is why RIS sponsored a shark-tank competition at the recent Retail Executive Summit.