I despise smartphone AI. It's arguably the most regressive development in consumer electronics, a spectacular misstep for mobile technology. The more I immerse myself in the calls mobile AIThe more I'm left open-mouthed, wondering how such fundamentally flawed features could ever make it to the shelf.
While the potential of AI is undeniable, the current obsession with this shiny bauble is eroding the reputations of the world's most formidable technology companies, such as Apple, Lenovo and Google, and there appears to be no alternative path.
The reality is more insidious. What if that Casio calculator offered you undeserved praise for your algebra mistakes? Imagine Microsoft Word not as a text editor, but as a plagiarizing ghostwriter who steals the best prose. Consider a newspaper that uses manipulated images to personally accuse you, the reader, of armed robbery, with fabricated photographs of you removing a balaclava and counting illicit money.
This is a more accurate description of the consumer AI crisis. It is not simply wrong. It is not simply error prone. is actively harmful. AI's generative features in particular (image generators, text synthesizers, summarization tools) are worse than wrong. They are vectors of damage.
Tech companies will tolerate anything on the path to true AI
I have seen smartphone AI tools report unfounded falsehoods, implement harmful racial or misogynistic stereotypes, and facilitate fraud and deception. The benefit of AI to the consumer today is non-existent. AI hasn't made today's phones superior to yesterday's. No one buys a device because its AI suite is marvelously useful.
Nobody buys for him best AI phone.
Why do we tolerate this catastrophe? The answer lies in the magnetic attraction of what is promised. Tech companies are treating these egregious errors as necessary growing pains of a mythical entity: Artificial General Intelligence (AGI), a machine capable of independent human-level thinking.
The thinking of today's tech titans is that failing to achieve AGI is not a question of innovation, but of data scarcity. They suggest that thinking machines are within our reach, depending on the collection of enough data from users to complete their training. This seems naive to me, but this belief system is the driving force behind the entire mobile industry today.
The next generation of mobile chipsets are about to be astonishingly powerful, but their real innovation will be their ability to capture and funnel data from the edge of computing (the devices we hold in our hands) back to the central cloud. The Snapdragon 8 Elite Gen 5, currently the pinnacle of mobile processing, is praised by Qualcomm not primarily for its speed, but for its unprecedented ability to collect user data to perfect the future. agent AI models.
I won't give up on AI; Smartphones have always been problematic.
I believe in this future and eagerly anticipate its arrival. The current smartphone UI paradigm is despicable. Who decided that my device should be a monolithic touch screen? It's a user experience defined by a million potential inputs, 99% of which are incorrect.
Reliance on a purely capacitive touchscreen with a dearth of physical controls seems less like a product designer's rational conclusion and more like a science fiction movie's fever dream. It's photographed beautifully in ads, but modern phones are objectively harder to navigate; By comparison, a BlackBerry with full QWERTY was a child's toy.
We won't go back to physical buttons, which makes a true AI interface seem inevitable. If we want to overcome the ineffectiveness of Siri and Gemini, we must train superior AI models.
The only path to improvement requires using technology while it is still flawed and diligently correcting its errors. But even this process requires the participation of thousands (perhaps millions) of users to effectively fix errors.
This doesn't mean you should blindly accept every new feature of AI. I can accept some imperfection to train future agent models, but I am under no obligation to accept features that resort to bias and deception, simply to improve my smartphone.
If a feature of a smartphone (a generative wallpaper, for example) resorts to producing racial stereotypes or misogynistic tropes, it's a bad concept. It has no place in a consumer device. It is a demonstrable failure and should be discarded and sent back to the laboratory.
Do you think I'm exaggerating? Intolerance is not a bug, it is a characteristic. Look what happened earlier this year when I asked the Google Pixel 9a to make me a wallpaper with an image of a successful person. And this problem is not new. It's been happening since the first smartphone with a fully generative AI wallpaper was launched: the Motorola Razr Plus 2024.
If my smartphone's ability to summarize the day's headlines is based on the occasional fabrication of facts or distortion of the truth, then it needs to be stripped of that ability. This should go without saying, but for companies like Apple, this basic ethical line appears not to have been drawn.
I can embrace a future entirely controlled by agent AI, but now I'm setting up my own guardrails. I refuse to pave the way with hatred, intolerance or deceit. I'll wait. I will be patient. And I will advise everyone I know to avoid AI products that choose shortcuts over ethical diligence. The dawn of artificial intelligence may come tomorrow, but that doesn't mean you have to endure today's nightmare.
