The Whiplash of Unboxing AI
The future of AI is coming upon us faster than we know how to process, let alone legislate. The whiplash is real.
PUBLISHED 5 JULY 2026
BY: Michelle Wronski
It feels like it’s everywhere, all at once, all the time. “Not sure? Just ask AI.”
Threaded through dinner conversations, morning news headlines, weekly team meetings. AI for this, AI for that. Customer service bots, personal assistants, therapists, tutors… an endless list of AI-powered everything. Get it? Got it? Good.
The rhythm is relentless. A new model, then a new bot. Each arrival lands for a few months before the next takes its place. What felt remarkable six months ago dissolves into today’s standard, and somewhere in that acceleration, it becomes harder to see where novelty ends and the new normal begins. If you asked me if I was thinking about this seven years ago, the answer would have been: “Yes, it's coming… but we have lots of time before AI is integral to our daily life.” Little did I know how quickly that future would actually arrive.
A Different Technological Era
A few dozen frontier AI models, seven iPhone generations, and one pandemic ago, the year was 2019. Before remote work became a mainstream option, before generative AI, before mass layoffs swept Big Tech, and before the race to build ever-larger models became the industry’s defining obsession, I, like many university students, was trying to figure out what I wanted to do with my life.
I had spent the previous four years working as a social media assistant whilst completing my bachelor’s degree. I watched follower counts skyrocket on clients’ Facebook pages, Twitter feeds, and YouTube channels in ways I could never have imagined before. Just a few years earlier, the idea of pitching commercial content on Facebook to my clients was absurd. As my then-manager put it best, “Who would go on Facebook looking to buy a million dollar home?” This was around the time when the Cambridge Analytica Scandal made headlines, a time when the scale of social media felt complex to grasp, like an uncontrolled beast increasingly difficult to ignore. My eyes could not look away from these developments, which led me further into the endless pit of tech policy.
After a long marathon of in-person university exams to wrap up the autumn term, I found myself on the couch in my apartment watching the YouTube series The Age of AI. On my TV, Robert Downey Jr. was explaining, with optimism, how it took roughly 10,000 years to go from writing to the printing press but only 500 more years to get to email, and that we were entering the dawn of a new age. I was well aware that AI was already quietly embedded in recommendation algorithms, shaping social media feeds, and on special occasions; the topic would creep up into my university lectures.
I remember watching the endless use cases unfold in the show: BabyX, a lifelike digital baby that learns and reacts like a human; Project Euphonia, helping people with speech impairments communicate; AI-powered prosthetic limbs; scripts written by machines for films; AI-powered pizza-making; robotic companions. I remember the double edged feeling of curiosity and absurdity creeping up inside me watching these product demos. Do we really want robotic companions? How would they add value?
These products didn’t feel like glimpses of the near future, more like ideas tucked away in pockets of Silicon Valley, removed from everyday life, a digital divide if you will. I turned off the TV, the projects slipping away from the screen and out of mind. At best, these seemed like nerdy late-night conversations, yet at the same time, somewhere in an office in San Francisco, OpenAI was quietly cooking away at GPT-2.
The Dawn of a New Era
It wasn’t until November 2022 when it felt like the starting gun went off. Not for the invention of AI, but for the public’s newfound interest in it. I was in the midst of choosing my master’s dissertation topic, meanwhile, ChatGPT was surpassing one million users in just five days. The EU AI Act was still in progress, moving through early drafts that didn’t account for general-purpose systems like ChatGPT, and policymakers were scrambling to keep pace. The conversations repeated themselves at each mention of AI: we need guardrails, responsible oversight, accountability, transparency, governance, and standards.
I decided my thesis would examine how small to medium-sized businesses attribute responsibility to AI, inspired by social media’s content moderation efforts and the increased attention the topic was getting at the time. The buzz around AI responsibility persisted; part of examining my research question meant finding use cases where AI was irresponsible, and what policies existed to manage them. These examples weren’t obvious or formally documented, but as my dissertation began to take shape, more examples continued to emerge around the world. These case studies, combined with interviews with business professionals unpacking definitions of “AI accountability”and “AI liability,”felt like organizing needles in a haystack.
On the surface of this research, I’d find funny stories, like an AI asking a journalist to leave his wife because the bot fell in love… odd, but not seemingly harmful. Yet the deeper you went, the more troubling the stories became. How do you manage a chatbot meant to connect you with emergency services in a life and death situation when it doesn’t? What happens if your mental health reaches the point that a bot convinces you to take your own life? How do you press charges when a bot convinces an individual they’re in love and actively encourages them to commit a crime like attempting to kill the Queen? These are real life documented examples, not fictional tales; situations people have found themselves in with AI that certainly didn’t cross my mind on that couch seven years ago. Who do you even point to that is responsible? On what legal basis could responsibility be attributed to this technology? Product liability? Negligence? Data protection?
At the time, there were no AI-specific legal frameworks to enforce such responsibility. Today, conversations about responsible AI appear more unified on the basis of shared vocabulary, but in truth, different organizations diverge in how those definitions translate into daily practice, and much work still remains to be done.As I am writing this article, a landmark US case has just concluded in which a jury found Meta and YouTube liable for designing platform features that contribute to user addiction and related harms. News outlets reporting Meta was assigned 70% of the responsibility for damages relating to issues including suicidal ideation, body dysmorphia, and attention deficits, while YouTube was attributed the remaining 30%. The lawsuit had originally included Snapchat and TikTok, but both companies reached settlements before the trial was set to begin. It’s a reminder of the similar strategies we saw being used against Big Tobacco, where the warning signs were in clear sight, but action lagged behind the harm.
This isn’t a clear-cut declaration that AI and social media are all bad or that we shouldn’t use it. We’ve found ourselves at a moment in time where changes happen faster than anticipated at a massive scale, and before we even have time to react and respond thoughtfully, more technologies are unboxed and widely distributed, so we don’t fully know what we're getting ourselves into. When we still had many questions about the impact of social media, AI entered the public sphere, pushing us to shift focus to the latest development. When I wrote that dissertation just three years ago, the conversations around AI liability, in its framing, felt like quiet whispers in a room of blaring music, leaving me with more questions than answers when I entered the workforce.
What now feels like whiplash is, in reality, a delayed recognition of the sheer scale and speed at which AI has entered our lives, with many of those questions still unresolved. Looking back at the girl sitting on that couch, what feels striking now is not only how much has changed, but how much was already there on that screen. The foundations of the AI we work with today were already in place, from the vast amounts of data used to train it to the systems designed to scale its output. Yet, in hindsight, it is only becoming clear how oversimplified the concerns we’re now rushing to solve once seemed.
So, where do we go from here?
As a new age unfolds, I can’t say that I have a five-step plan to address the AI-induced whiplash, or that there’s a single path we need to follow. Perhaps this is instead a reminder that, regardless of technical background, we all have agency in how we shape and use these tools as individuals just as much as we have agency in how they shape our lives. Asking questions matters. Shared understanding matters. And the more we align on what we want, and what we don’t, the greater our ability to shape this technology in ways that benefit all of us, far more than we might expect. Understanding comes gradually, in fragments of time, shaped by the choices we make. Perhaps that is the work ahead: not to resolve all the uncertainty, but to engage in the conversation about what we’re building together.
Michelle Wronski is a London-based researcher specialising in technology policy and its societal impact.