Didn’t “Just Do It”: When AI Runs, But Humans trip over potholes

STEP 1: My Notes and Thoughts 👟 Nike- to “just do it” or not to “just do it”, that is the question.

I can’t help being wary. Who really owns the data in all this? The customer? Nike? Or Cognizant, (their partner running global tech operations under that five-year deal)? What happens after those five years does somebody else get our data? With so many players involved, it feels like too many cooks in the kitchen playing fast a freely handing over MY information to eachother, and the recipe is people’s vulnerable data. Nike talks a lot about sustainability, but I wonder how sustainable it really is to keep feeding energy-hungry AI systems in the name of efficiency. They’re chasing innovation, sure, but at what environmental and ethical cost? My Husband wanted to do the Hoka sneaker experience as everyone and their mother did it when it came out only several friends said they ended up throwing them out because even if they were made to AI measure for them, they seemed to miss the mark on things we unexpectedly come across in runs that will effect the use of our shoes. HOKA’s integration of AI and biometric data is marketed as “enhancing the human experience” but it’s still based on generalizing from massive datasets. That means some outlier feet (unusual gait patterns, disabilities, prosthetics, pothole like mine fields etc.) might not fit into the algorithmic logic.

As O’Neil (2016) says in Weapons of Math Destruction, models trained on majority data can unintentionally exclude minorities or exceptional cases. That’s why in future iterations, companies like HOKA will need to ensure inclusivity in biometric modeling not just performance optimization for the “average” runner.

👗Amazon HA!, I currently have a package waiting to be returned because the “AI size recommendation” didn’t work, so my confidence level is low. I also think the algorithm struggles to understand that one account might shop for multiple people in my case, family and friends. That alone must confuse the data. And if Amazon ever asked, “Who is this swimsuit for?”, I’d delete the app immediately. That level of personal questioning crosses into invasive territory. I’ve never handed over my body data no BMI, no scans, so unless Amazon starts scanning customers, I don’t see how their system can really deliver the accuracy it promises.

💳Mastercard Frustratingly I’ve been on the wrong side of Mastercard’s AI fraud detection more times than I’d like. My card’s been blocked for totally normal purchases, and it’s incredible it didn't seem to learn from it! It happened several times! I understand the logic stopping fraud before it happens but it often feels like the system is blocking legitimate customers for their own money. AI can be great at pattern recognition, but not so great at recognizing context. Sometimes, being “protected” feels more like being policed.

🧠 Common Thoughts

Across all three cases Nike, Amazon, and Mastercard I see the same tension between helpfulness and intrusion. Each company claims to empower customers, but in practice, they collect immense amounts of personal data with limited transparency about who controls it. These systems promise personalization and safety but often end up reducing human nuance to machine logic. AI might make our lives easier, but it also makes me wonder: at what point do we stop being customers and start being data sources?

Step 2 ): after reading Mathew’s post. I worry…When AI Thinks in Straight Lines but Humans Don’t?!? The more I study cases like Nike, Amazon, and Mastercard, or experience AI myself, the more I realize AI’s biggest weakness isn’t a lack of data, it’s a lack of nuance. These systems are designed to optimize patterns, but real life doesn’t always fit the pattern, but humans are exceptional there will always be new and creative exceptions to every rule.

Relying purely on AI for decision-making feels a lot like being trapped in bureaucratic red tape: the rules look logical on paper but collapse in context especially if your an outlier. Mastercard’s fraud algorithms flagging ordinary purchases, or Amazon’s size recommendation system failing anyone who doesn’t match its data profile. These aren’t glitches; they’re symptoms of a deeper issue in that AI’s unable to process exceptionality (O’Neil, 2016).

As Bryson (2020) argues, machine intelligence lacks the adaptive empathy that makes human governance flexible. Societies evolve, cultures shift, and values change but algorithms, once deployed, tend to fossilize the assumptions they were trained on. Like outdated laws, they can unintentionally marginalize the people they were built to serve?

If organizations continue to treat AI as an authority rather than a tool to review and check before sending to Clients, we risk replacing human judgment with automated bureaucracy… hahah only this time, it will run at the speed of data.

Funny enough We are not perfect so the challenge isn’t to make AI perfect, it’s to make sure we don’t lose the imperfections that make human systems resilient, inclusive, and capable of growth.

The blog entry was edited on 10 November 2025 at 10:24 PM, as displayed in KEATS. Personal experiences and academic reflections are presented in their original context.