Subscribe to discover Om’s fresh perspectives on the present and future.
Why Tech Needs Personalization
Every weekend, my good buddy Hiten Shah and I head to The Coffee Movement’s Balboa Street location. They make a nice pour-over. And while it’s about a half-hour from my apartment, the coffee is worth the drive, the time, and the money. It’s a long enough drive for us to talk about life, work, and, most importantly, new ideas about technology. Sometimes, if our schedules sync up, we even go twice a week. What doesn’t change is where we meet. It’s the same location. We drive the same route. The only variable is the time of day we go for our coffee.
On the way to the café recently, Hiten switched the car into self-driving mode. Tesla does a remarkably good job pilotingitself, especially on crowded streets. It’s far more careful and attentive than some of my fellow human drivers. As we drove, I wondered aloud: Why can’t Tesla learn our favored route? Why can’t it pick up our weekly driving patterns and store that information in the system? It doesn’t have to upload anything to the cloud; it could just keep it local. Whenever it detects the start and destination, it should automatically follow our preferred route, adjusting only for traffic.
Surely, there’s a way to add a personalization layer that makes maps more effective. And why stop with Tesla? So why can’t Apple Maps or Google Maps do it? This shouldn’t be too difficult. Unless there’s some regulatory reason they’re not permitted to do so.
Personalization isn’t a feature; it is a way of acknowledging that life has texture. My own thinking around this is no different from Neil Postman’s, who, in Technopoly, pointed out that “technological change is not additive; it is ecological.” I’ve written before that data without empathy and understanding is meaningless. The efficiency and convenience of technology are what deaden our human experience. Without personalization, we’re reduced to mere data entries.
Personalized mapping looked like the future when Google bought Waze; yet Google Maps remains only somewhat personalized. It is rarely enough to be genuinely useful. And Apple, as a company, gets a failing grade even on the most basic personalization, so we shouldn’t expect much from Apple Maps.
I’m trying to highlight a core tension between algorithmic efficiency and user personalization. Mapping systems often funnel users onto algorithmically determined routes that prioritize speed and major roadways. I suppose this is a deliberate design choice. These algorithms are optimized for the lowest common denominator: moving the greatest number of people to their destinations as quickly as possible, typically via routes the system deems most efficient.
I recently met Rodney Brooks, one of the giants of AI and robotics. He shared a little story from his ride over to see me:“I came here in an Uber this morning,” he told me, “and I asked the guy, at one point, what street we were on. He had no clue. He said, ‘I just follow it.’” (Read my interview with Rodney Brooks over on CrazyStupidTech.com.)
We’ve already become machine-idiots, following our machines blindly. We keep taking the routes handed to us by mapping systems, and we’ve stopped even trying to find our own way. No wonder the algorithms jerk us around.
I’m often confounded when Uber drivers take freeway detours, even when city streets would be faster. Lacking local street knowledge, they inadvertently reinforce the system’s biases, feeding it more of the same data it then uses to direct future users. With deeper, more contextual understanding of real-world scenarios and user intent, that wouldn’t happen; we’d move beyond simply adhering to a prescribed, albeit “fastest,” route.
Despite increasingly sophisticated edge computing in our devices and vehicles, true personalization remains elusive, pointing to more fundamental limits. Even today’s artificial intelligence systems are hampered by this absence.
“We shape our tools, and thereafter our tools shape us,” John Culkin once observed. In our post-algorithmic age, theopposite seems true: the tools built to serve us end up reshaping us. In our pursuit of efficiency, we’ve ceded the subtlety of choice; the small human quirks that make each journey ours are sacrificed to the machine idiocy we’ve come to call convenience.
No one put it better than the late Steve Jobs.
“Technology alone is not enough — it’s technology married with liberal arts, married with the humanities, that yields us the results that make our hearts sing.”
Personalization might seem, at first, like just a technology feature. Yet today’s maps don’t recognize the weekly coffee shop trip as ritual, and algorithms don’t sense the rhythm of a drive shared between friends.
It reflects who we are, and it’s an opportunity to make technology and machine intelligence more human, more personal. Until our systems understand that, they’ll keep optimizing our lives into neat, predictable boxes even as the messier, more meaningful routes go unlived.
October 29, 2025. San Francisco
Editor’s Note: It is a sign of the times that a substantial number of you are taking advantage of newsletters to read what I publish. I have made a subtle change to the website: we are now publishing as a “newsletter.” If you are a subscriber, nothing will change. If you are not, please consider subscribing — and tell your friends and colleagues to do the same. In doing so, you will be supporting my writing. My goal is to publish up to three times a week — including a weekly list of my reading recommendations. (RSS feed is unchanged.)
Are you sure you want this? Facebook is personalized.
I would argue that Facebook is not personalized at all. It is an illusion of personalization by algorithm meant to see you ads and not to really give you a meaningful social experience. That really is the point of my piece— none of these companies actually allow us to add layer of self. They do everything for themselves.
Thanks for the newsletter Why Tech Needs Personalization.
Very thought provoking…I will need to think deeply about this for awhile but you have confirmed/clarified something I have been feeling for quite awhile.
Thank you!
I would love to know what your conclusion is, so do come back and share your thoughts.
A “personalization layer” would seem to be an obvious opportunity for a company like Apple. Build it into the client, and have it communicate with the cloud. What would we pay for it?
Also, I remember 10 years ago, visiting my late mother, I rented a car at LAX and plugged in Waze, asking for a route to Huntington Beach. There was an accident and it ran me right through Compton, including some nasty intersections where I had a stop sign and the main road didn’t. Harder to get those options now.
I remember you telling me that story at some point in the past. Whether on the blog, or via email. And to think we had MyYahoo in 2000. It has taken this long for “nothing to happen” 🙂
So true.
I live in Scotland, yet the number of times I search Apple Maps for a place and it presumes I want somewhere in the US, Google Maps does it too. It’s ridiculous and SURELY easily fixed?
Apple has a tendency of being over aggressive about privacy settings. You might need to re-enable location access to the maps and make it “all time” access instead of “when app in use.” Apple is not a very smart about trying to make software good and useful these days — too many settings and too much effort on our part 🙂
Indeed, and I have been through every setting, and combination of settings to try and rectify but to no avail.
Driving directions to Perth? (About 75min drive) and I get the one in Australia….
Personalisation would be nice but first get the basic right. At least Google Maps knows which country I’m in.
But yes this surely must be the next thing to “fix”. AI isn’t too bad at taking and storing info about me, so when I ask it for recommendations for a bike it doesn’t list ladies or kids and specifically finds me XL frames.
Apple and Google have MORE info about me but fail this stuff so quickly it hurts.
Apple (and its machine learning) is a different class of incompetent. They mean well, just don’t know how to execute rightly on that. But then they have to serve the lowest common denominator and thus make convenient decisions.
Hi Om:
Here, here. I have seen the same idiotic directions from Waze, though it does know which days on which I go to the Rossi pool or the Sava pool for my morning swim.
Cheers,
Chaz
Would love to offer another lens on your insight:
Perils of Cognitive Offloading – The AI personalization would be equivalent to a combined Lazy Boy, TV and Inexhaustible supply of cheese puffs for a human. In short, we’d love it so much that we would forget how to remember things. Text messaging led to “hyper co-ordination” and the end of “I’ll be there on time” versus the rise of ” I’ll be there in 10 there was an xyz on the floor of my Uber, running behind.” Or my Vice? I get Alexa to remind me and still forget to ACT on it (palm face)
https://www.reddit.com/r/Futurism/comments/1ihktwr/cognitive_offloading_the_future_of_human_evolution/
My point: The brain is a complete -bas***d-.
It is lazy. Give it any excuse to get it into low power mode and it will take oh so more. We need a cultural drill seargeant to whip our collective wills into shape, not “hold space” for weaponized technologies of human entrainment presented as entertainment.
So many technologies were invented without the unforseen consequences. Again, the brain…
Still. This old and lazy brain is so in love with AI. Especially how AI allows me to think faster and forget more useless stuff. It used to be important that you knew something: facts, figures, order etc. Games changed.
WAY more important that we now know WHAT TO DO WITH THINGS QUICKLY post AI. Those will succeed. Systems thinkers and strategists will succeed. “Crafters” will succeed.
Thanks Om. You are very right about mapping systems. In Israel we use mainly Waze, and I noticed that in some situations, Waze recommends a specific route, and since there are thousands of users who follow the recommendations, that route becomes a bottleneck
This is why we need systems to become adaptable and personable. I wish more companies paid attention to these little things.
I’ve been known to drive clear across town (Seattle) for good pour over coffee. Worth the expense and time.
A man after my own heart, though will you drive to San Francisco for a great cuppa?
Hi Om, I’m responding after a long time… This really resonated with me. We’ve traded personalisation for efficiency and lost something real in the process.
You nailed it -simple and personal should matter more. And yeah, social media is the worst offender. It’s supposed to bring us together, but it just makes everything feel hollow and addictive.
What gets me is how we’ve stopped thinking for ourselves. We just follow whatever route the GPS tells us, without question. We don’t even know the streets anymore. The algorithms don’t care that your coffee run isn’t just about getting coffee—it’s ritual, it’s time with a friend.
Technology should fit into our lives, not the other way around. But instead, it’s reshaping us to fit what it wants.
Thanks for this. We need more people saying this stuff out loud.
Thank you for your insight. Algorithms lacks soul.
Spot on!
I have noted with some dismay that young people who came of age in the era of pervasive digital cartography have failed to develop any ability for planning using global geodesy. I’be been asking people to describe how they navigate the regional road network. In particular, when needing an alternative route to a destination, do they visualize a regional map at some resolution, or do something else.
That usually requires an explanation. The most general approach offered is a path based on visual memory of the interconnection of real intersections and exit landmarks. This explains the significant backtracking triggered by a transient disruption in the expected path. The driver must use large-scale landmarks to get back to an existing path to the desired destination. If,
however, there is no existing known path, planning resorts to an arbitrarily coarse regional approximation sequence using well-known arterial paths until intersection sequencing can rake over. This means that a very small surprise local diversion can easily result in a kilometer-scale recovery path. Localized alternate path density cannot be used for localized recovery because landmarks have not been seen before. Cheap speculative localized search cannot be used because local knowledge, real or assumed, does not exist in the mental images
It is true that in completely unfamiliar territory, the large-scale landmark matching approach works for macro networks down to some scale. But when the scale drops to “surface streets”, the inability to use small-scale maps is a serious problem.
When the map makers realize they can charge by drink, “grave disorder will result.”
So for Holyday Season this year, when buying for a still-impressionable young mind, consider giving real, paper maps of their home area. (The Thomas map books are great!)
Then spend some time helping them to learn how to exploit information not rendered in text.