AI Initiatives at the US Post Office: Final Rethink Before We Dissolve It!

MIT released a comical study in May of 2020 on the historical innovation accomplishments, and the potential logistical endeavors brought to light by the beleaguered United States Postal Service. The apparent scholarly article cited several technological implementations and employee empowered frameworks that could define the USPS’s plight over the next five years. These frameworks, through FY2025, mention a postal-banking initiative and two AI autonomous vehicles initiatives. These initiatives are, unfortunately, compulsory and will fail, wasting hundreds of millions in taxpayer dollars. Despite its popularity in some rural parts of the country, the USPS has struggled with declining revenues and increasing costs over the last two decades, recording $70 Billion in cumulative losses since 2007. It secures mail delivery to 160M addresses and employs approximately 600,000 employees.

As USPS revenue continues to decline, operating costs continue to rise, and personnel costs account for 76% of the postal service expenses. Much of the aging delivery vehicle fleet, which was acquired as far back as 1987, include costs of about $4500 per vehicle per year in maintenance costs spanning more than 250,000 vehicles. Further, postal vehicles were involved in over 100,000 accidents since 2015. In 2018 alone, postal vehicles were involved in more and 30,000 accidents, and in 2017, 11 postal workers were killed in roadway accidents. In light of its uncertain financial status, the Postal Service has been designated a high-risk agency by the GAO since 2009.

Currently, innovative companies in both public and private sectors attempt to future-proof their organizations with AI initiatives to stay competitive. Before diving in on why it is time to dissolve the USPS, a quick chronological note about how pioneering the post office has been historically-from delivering mail on horseback in the 1700s to utilizing swift steamboats in the 1800s to airmail and, more recently, the notion of drone delivery over the next couple of years.

Yikes! Can you imagine drone delivery by the USPS? As you well know, Amazon received an FAA 135 certificate for drone deliveries, and the USPS has requested an RFP for UAVs earlier this year. Drone delivery by USPS is as frightening as going to the mailbox to sort through the voluminous coupons from merchants we do not care about doing business with. To that end, there is still no proof of concept from the USPS stating what the future of mail and package delivery looks like. Additionally, there is no recent historical reference of innovation other than Informed Delivery, which can provide a glimpse of what an efficient logistical solution looks like when delivering packages that can match the private sector.

In August of 2020, the U.S. House passed a bill to provide the USPS with another $25B in funding. It has no chance of passing in the Senate, however. Despite the USPS’s “universal service obligation,” which requires it to maintain postal facilities and personnel to serve every zip-code in the country, we must hurry and dissolve the USPS as quickly as possible.

Survey Results from AI Initiatives and Measuring its Impact

Because of the black-box nature of AI in government settings, and lack of explainable AI programs at the USPS, technology teams must address genuine understanding and engagement to the end-user. In two recent surveys concerning the AI initiatives by the USPS, a high-level overview of the results show that many employees have expressed the reasons behind the delays in incorporating AI initiatives, based on a lack of understanding of the models. The Center for Stanford University’s Human Center for Artificial Intelligence wrote a post two weeks ago, on the challenges of measuring AI’s impact in government agencies.

According to the survey, acceptance and adoption were suppressed when the solution seemed too complicated or not understood by the end-users (mail carriers). This is merely one core reason why AI will not manifest efficiently at the post office. For successful AI implementations, everyone needs to be on board and advocate for the same initiative. Internal dissenters can sabotage AI projects. Second, there is a limited knowledge base within independent governmental agencies, including scarce human capital, to substantiate a formidable AI initiative; without the help from the private sector. To restructure the USPS through “explainable AI,” it must rationalize the project and win approval specifically from the 600K employees.

As it relates to the AI use case for autonomous vehicles and autonomous long-haul trucks, public opinion from a nationwide poll was not favorable as well. Polls suggest that many believe autonomous vehicles might limit traffic incidents. Still, the poll also expressed that the public might not believe that the USPS can be trusted with autonomous Delivery, much like Drone Delivery. Further, among the four major logistics companies in the US, (Amazon, UPS, FedEx, and USPS), the postal service was rated last. The entity which conducted the survey, OIG, also released a statement saying that public perception must be monitored as part of its AI deployment strategy. If public perception is not changed, the autonomous vehicle project and other initiatives will have vanished.

Private Banking and Real Estate

Americans are awakening to both the proposed value proposition of the USPS and the inherent dangers of privatization. The crisis has also sparked renewed interest in postal banking; proponents believe it is a win-win approach that could make the USPS more financially resilient and provide badly needed financial services to tens of millions of underbanked Americans. That is hogwash.

Several dozen private banking apps are looking to serve the underbanked, and Paypal is one of them. So this argument is a non-starter. One American bank has considered placing an ATM in each of the 30,000 post office locations. This would be a shortsighted aesthetic marketing move by the bank brand to become a channel partner with the fledgling post office. First, no-one uses cash anymore, and second, it would damage the bank brand, given the fragility of the USPS to sustain its offices or services.

The post office has more than 30,000 locations. When considering dissolving the independent agency, one question comes to mind; what to do with the real-estate? While some of the mail sorting facilities that are 200-400K square feet, converting them to shipping warehouses operated by private companies can mitigate some of the costs associated with dissolvement. Companies such as Prologis, Amazon, Walmart will initially devour these strategic locations. For the public post offices themselves, converting them to FedEx Locations and UPS Store locations is not a far-fetched idea. Currently, there are 7100 FedEx Office and UPS Store locations, and a bank partnering with these companies would prove more strategic.

Moreover, private enterprises have developed unmatchable logistical efficiencies for package delivery that make it seamless to receive and return items purchased. The Amazon/UPS Store collaboration has made returns convenient (whereby packaging is not required) by an end-user, and the refund is immediate with a QR scan. So, there is some light at the end of the tunnel in this regard when repurposing the real-estate.

Many scholars believe that the postal service can pivot during this super-cycle of innovation. However, I believe that they cannot- not only in the competitive nature of logistical execution by private companies but also the regulatory hurdles that will thwart streamlining the initiatives.

AI Use-Cases in an Uncertain Regulatory Future

In traditional technology cycles, one is given the input and creates an algorithm that transforms the input into the predictable output. With one machine learning method, the algorithm or logic is developed by the computer, mostly training a dataset using unsupervised learning, leaving it more difficult to interpret or explain. Once again, this will inhibit end-users at the USPS embrace any project, without a thorough understanding. This is also a point of contention with many private companies with human capital juxtaposing different machine learning models to figure out ways to communicate the output more simplistically.

Secondly, the idea of using autonomous vehicles in rural areas while the mail carrier sorts through the mail in between stops is ridiculous and too simplistic. Sure, this may cut down on traffic incidents and vehicle maintenance costs and save lives, but it is not a fix-all for the government agency, which is bloated by governmental regulation. There are also legal implementations that relate to the use of autonomous vehicles. In early 2020, congress and state legislatures have been figuring out ways to regulate autonomous vehicles’ manufacturing rather than increasing their use. The most immediate legal implication relates to the Postal Services tort liability for vehicle accidents. Although autonomous vehicles are projected to be more efficient and reduce accident frequency, they will not eliminate collisions as long as human drivers share the road.

Informed Delivery has 17 Million Subscribers

The only positive advancement of the postal service in the last couple of years has been their Informed Delivery Service, which provides an image scan of your first-class mail. It is sent to you daily by email and alerts you to what type of mail is coming to your physical mailbox. It currently has about 17 million subscribers and boasts an open rate of 88%, which is quite remarkable. That demonstrates that people are still connected to their first-class mail. Nevertheless, this represents only 10% of the addresses in the U.S.

Responsible AI indicates that people who deploy AI must use appropriate judgment and be accountable for solutions produced. Equitable AI indicates that bias and discrimination should be avoided and monitored. Traceable AI should be transparent and auditable. It would be challenging for USPS, based on general public opinion, to deliver on all of these initiatives in the foreseeable future.

There is no way to future-proof the USPS and without explainable AI, not only to their 600K employees but to the 160M addresses they serve. Their current initiatives are bloated with regulation and not specific enough to discern. While the USPS believe they are shaping the future of work, the competition from other logistics companies is too great. Uber Freight just received 500M in funding, Amazon, FedEx, and UPS, along with many others like XPO logistics, are far more advanced. Imagine, if you were living in China, would they even consider a government entity so poorly run.

Latest posts