Examining The Interdependence of AI and RFID

Published: February 26, 2024

Artificial intelligence and RFID are becoming increasingly co-dependent as AI enhances the data RFID provides its users, while RFID feeds AI with the data it needs train itself.

This story is the first in a series about what AI means for the RFID and IoT industries. 

It’s been nearly a century since the public was first introduced to artificial intelligence (AI) in the 1927 dystopian movie Metropolis. Today, the promise of AI performing tasks and delivering information that long relied on manual efforts seems limitless.

Make no mistake: from personalized shopping to automated video creation to fraud prevention, the learning power of computers is already making giant strides that are impacting our daily lives.

For the RFID and IoT industries, AI is viewed as both potentially a solution and an opportunity.

The Solution: AI enables RFID technology deployments to provide functionality that was out of reach in the past.

The Opportunity: any AI system demands data—lots of it—to begin training itself, and RFID provides that with each tag read.

Despite marketing proclamations, AI does not solve all the ills of the world. The RFIDJournal.com over the last couple of months talked to  experts in the industry to help us unpack what AI will and won’t do for the RFID industry. This is the first in a series examining AI in the world of RFID and IoT, as well as related technologies.

RFID Journal Live

AI Creates New Promises for RFID

Those that we talked to agree that AI is poised to be an accelerator in RFID adoption. A decade ago, RFID technology concerns were centered on whether, and how effectively, the hardware works. The industry has moved past many of these fundamental concerns.

The other side of the coin has been data management. Some technology companies in the past were lamenting that RFID systems weren’t being used to their capacity. The question was, how are users benefitting from, or managing, the “big data” that RFID produces.

“The potential of AI is real and is happening today, but we’re in the very, very early days when it comes to RFID,” said Sandeep Unni, Gartner’s senior director analyst for retail industry practice.

Since RFID generates an extensive treasure trove of data, Unni said, the value for AI is in the analytics and insights it can generate from this data.

What AI Offers RFID Solutions

Unni pointed to an example: “Upstream inventory planning systems can consume the real-time inventory data generated from item level RFID,” and when that data is then combined with AI, the software can unlock granular and prescriptive inventory optimization decisions.

“These inventory implications can be transformative,” Unni said.

Similarly, machine learning algorithms can be used to analyze item-level sales performance, customer engagement and conversion rates, and provide e-commerce like analytics in a retail store environment.

Applications like this benefit from AI’s ease of use. It can help process RFID data more efficiently, extracting the insights and patterns users can gain from, said Gus Rivera, Mojix’ CTO.

What AI Can’t Do

There are a variety of challenges AI is not designed to address. For instance, physical tagging still requires human intervention or automated machinery, Rivera pointed out. And users cannot expect AI to overcome hardware specifications of RFID technology.

“For example, it cannot extend the range of RFID tags beyond their designed capabilities,” Rivera said.

Perhaps most importantly, it cannot take the place of human thought and analysis, which means oversight will always be necessary.

“AI should be seen as a tool for informed decision-making rather than a replacement for human judgment,” Rivera said.

What AI Can Do

The value of AI depends on the quality of input data. If RFID data is inaccurate or incomplete, AI may generate unreliable results, Rivera pointed out. And he added that there may also be ethical and regulatory compliance issues that AI could create through its self-learning systems which cannot ensure ethical or regulatory compliance on their own.

What AI can do, at an unprecedented level, is simulate increasingly complex scenarios and outcomes, and its recommendations or automation go far beyond what human effort alone could accomplish.

“Today, we’re really modeling very real-word complex scenarios and running millions of what-if scenarios to determine optimize recommendations [and] actions,” said Rivera. “This is the culmination of IoT maturity, big data and cloud computing converging at the same time.”

As part of the effort to bring the benefits of AI to RFID, Mojix has been working with Google Cloud and their own AI team with access to data teams at the Google Analytics AI Summit.

The promise of AI is significant, Rivera stressed. Overall, AI’s impact on the RFID industry will be driven by its ability to harness the power of real-time, high-resolution, item data, and leverage AI for automating anomaly detection, simulating outcomes, decision-making processes and optimized operations.

Intelligent, Adaptive and Responsive

As AI technology continues to advance, it will enable RFID systems to become more intelligent, adaptive, and capable of delivering value across everything from logistics and retail to healthcare and manufacturing.

AI can help process and analyze the vast amount of data generated by RFID systems more efficiently and effectively. It can identify patterns, anomalies and insights that might be missed by traditional methods. At the same time, predictive maintenance algorithms can be used to monitor the condition of RFID tags and readers, leading to proactive maintenance and reducing downtime.

A real world example is the use of AI algorithms to analyze RFID tag reads, detect counterfeit products and prevent fraud within supply chains, especially in industries like pharmaceuticals and luxury goods.

Anomaly detection to optimize inventory levels through insights and recommendations (stock shortages, excess inventory, stock rebalancing) is a benefit AI offers now with RFID.

“Now we’re moving into traceability use cases with food safety as well as luxury/food compliance, regulatory use cases, and supplier score-carding,” said Rivera.

An Ounce of Caution

Not surprisingly, however the RFID industry is not immune to the AI hype that is underway in some forms across all industries.

“Rushing into any AI initiative or pilot for RFID without preparation is a sure shot way to stumble and fail,” Unni said. He advised to “start with clean data. The importance of a robust data foundation cannot be overstated.”

Additionally, claims made by companies in the AI industry should be approached with a healthy degree of skepticism and critical analysis, Rivera said. AI is a broad and evolving field, and it’s important to differentiate between what is currently possible and what is a future aspiration or exaggeration.

Downside of AI

There is always the potential for AI to introduce some negative results as well. Outside of RFID technology use, New York University’s Stern School of Business professor Hanna Halaburda indicated in research titled “How Artificial Intelligence is Shaping the Economy,” that AI can be used for collusion in dynamic pricing in the retail environment and could risk some health outcomes in the healthcare market if too much trust is put in the technology.

Beyond negative outcomes and despite claims by marketers, AI cannot guarantee absolute security, only enhance certain measures. Security in RFID systems depends on physical safeguards and best practices.

“AI’s predictive capabilities are probabilistic and based on historical data. It cannot provide perfect predictions in all situations, especially when dealing with unprecedented events,” Rivera said. “But it’s getting better every day.”

Early Gains for Retail

In the world of retail solutions, AI is targeting RFID data already in some stores to enable new ways to understand what is happening on sales floors. The benefit of using RFID for such systems is its innate privacy, said Sam Vise, CEO of retail software company Optimum Retailing.

RFID is blind, Vise pointed out. Feeding video streams to an AI system can yield results that mean individuals can be identified or categorized. RFID, on the other hand, tracks a tag, rather than a person.

Therefore, with RFID, Vise said, “we can see the movement of the product in the store, we can see it was purchased and we can see if it’s on display—we can get a lot of information while not infringing on anybody’s personal privacy.”

Looking ahead at the convergence of RFID and AI, Vise says, “I really think AI is going to improve that shopping experience” with the data provided by an RFID system.

RFID is one way to get the large volumes of data needed to start understanding patterns.

“It’s kind of our heads in the machine,” he pointed out. “We have to train it as humans, using the AI and the machine learning to try to pick up the overall patterns.”

Key Takeaways:
  • AI offers promise for the RFID industry, while the technology gains from the data that RFID can provide.
  • A level of caution will help RFID technology users ensure their AI solution works as anticipated and doesn’t create problems.