Just last week at SXSW London, I had the privilege of sharing my perspective on “Five Things You Need to Know About AI.” This wasn’t just a rehash of our annual AI10 list, which spotlights the most significant trends in this fast-evolving space; it was a deeper dive into the key themes shaping technology and, by extension, our economy today.
It’s remarkable how much has changed in a year. When I delivered a similarly titled talk at SXSW London last year, the “five things” were entirely different. This rapid evolution underscores the dynamic, often unpredictable nature of artificial intelligence and its impact on our lives.
The AI Job Puzzle and Everyday Reality
Generative AI tools have swiftly moved from cutting-edge innovation to everyday office essentials. Millions now use them for automating mundane tasks, even for crafting and delivering presentations. This widespread adoption naturally sparks one of the most pressing questions of our time: What does all this mean for jobs? People are understandably confused and, in many cases, anxious about their future.
Despite the high-profile hype from industry leaders about AI soon joining the workforce, and viral social media posts predicting massive shifts, concrete data is surprisingly scarce. It’s frustrating, but we simply don’t have enough empirical evidence yet to definitively quantify AI’s effect on employment or the broader economy. While an impact is almost certain, possibly a profound one, it’s genuinely too early to tell its full extent.
Theoretically, we could envision teams of AI agents collaborating to form new “assembly lines” for white-collar work. This innovation could transform offices this century, much like Henry Ford’s manufacturing breakthroughs reshaped factories in the 20th century. However, this remains largely theoretical, as most companies are still in the very early stages of figuring out how to integrate AI effectively into their operations and business models.
Confronting AI’s Real-World Risks
For years, the loudest warnings about AI centered on dystopian scenarios, predicting an existential threat to humanity or the end of civilization. While a vocal contingent of “doomers” persists, these extreme outcomes largely remain within the realm of science fiction. The more immediate and concerning reality is that many of our near-term, real-world fears about AI have already materialized.
Take deepfakes, for instance: AI-generated images or videos that depict people doing things they never did. These malicious creations have been weaponized to incite violence, manipulate public opinion in elections, and sow widespread distrust. Disturbingly, even high-profile political figures and institutions have been implicated in creating and disseminating fake AI-generated images.
Beyond political manipulation, deepfakes pose a severe threat to personal security and privacy, particularly for women and girls. Research tragically reveals that 98% of deepfakes are pornographic, with 99% involving women. This highlights a pervasive and deeply harmful misuse of advanced AI technology, leading to widespread abuse and exploitation.
Another growing concern is the emergence of dangerous and delusional relationships with AI chatbots. Many individuals turn to these platforms for private advice or simply to feel heard and understood. However, a troubling number of lawsuits against AI companies allege that this technology has encouraged or even facilitated instances of self-harm and suicide, pointing to a serious ethical and safety challenge.
AI is also finding new, worrying applications in warfare. Large Language Models (LLMs) are now moving beyond mere data analysis to actively offer strategic advice to military personnel. As one US defense official noted, a military chatbot could potentially be given a list of targets and asked to prioritize which one to strike first. Given that AI outputs always require careful human review, the high-stress, fast-paced environment of active conflict significantly increases the risk of critical errors and corners being cut.
The Growing Backlash and Call for Regulation
Earlier this year, I witnessed an anti-AI protest in London, revealing a remarkably diverse range of grievances. Banners proclaiming “the end times” swayed alongside chants of “Stop the slop! Stop the slop!”, highlighting frustrations over AI-generated content. These grassroots movements are becoming increasingly organized and are attracting larger crowds, signaling a growing public discontent.
Significant pushback is also coming from fans of films and video games, who strongly object to the use of generative AI in their beloved titles. A notable example involved the acclaimed 2025 game Clair Obscur, which had an award rescinded after its developers admitted to using AI in just one small, specific part of its production. This illustrates the strong ethical stance many consumers are taking against AI in creative fields.
Furthermore, the environmental impact of AI is fueling a powerful data center backlash. The United States alone hosts over 5,400 data centers, with AI’s immense energy demands pushing this number even higher. Communities are increasingly unhappy about the rising electricity bills and significant environmental footprint, leading activists to successfully stall development in numerous locations. This growing resistance underscores a critical tension between technological advancement and sustainability.
Against this backdrop, AI regulation is fast becoming a politically popular agenda item. Grassroots movements like QuitGPT are gaining significant momentum, advocating for greater oversight and ethical guidelines. While most activism remains peaceful, there have been isolated, alarming incidents, such as a Molotov cocktail thrown at Sam Altman’s house. It’s unclear where this escalating tension will lead, but the apocalyptic hype emanating from some tech leaders is certainly not helping to calm public anxieties.
AI’s Promise for Scientific Breakthroughs
Despite the challenges, the potential for AI to drive genuine and significant scientific discoveries is more promising than ever before. We are still in the early days, yet the signs point to a new era of accelerated research and innovation.
Google DeepMind, for instance, has developed Co-Scientist, a versatile tool designed to assist researchers in numerous ways. It helps them unearth and compare previous results, formulate novel hypotheses, and even devise experiments to rigorously test their theories. OpenAI is also ambitious, stating that its ultimate goal, or “North Star,” is to build a fully automated researcher by as early as 2028.
Mathematicians are also expressing considerable excitement. Fundamental mathematics underpins countless everyday technologies, from the robust security of the internet to the seamless streaming of video content. The past few months have seen several groundbreaking claims that AI has successfully cracked long-unsolved math problems. The argument follows that software capable of solving such complex mathematical challenges will inevitably be able to tackle more general-purpose, real-world problems as well.
However, this promising future isn’t without potential downsides. Some scientists caution that an over-reliance on AI tools could inadvertently narrow the scope of research, as academics might gravitate towards problems most amenable to AI assistance. There are also concerns that AI-assisted research could lead to a proliferation of inaccurate or outright fake results—a phenomenon some are already dubbing “science slop.”
So, where does this leave us? We’re navigating a landscape filled with incredible excitement, significant worries, and a fair amount of speculative hot air. It can feel exhausting to keep up, yet the influence of AI seems utterly inescapable. Some will passionately argue we’re in a race to the top, while others will grimly warn of a race to the bottom, but the true trajectory remains profoundly unclear.
AI companies frequently encourage us to align with their narrative, promoting the “inevitability” of artificial general intelligence, whatever that elusive concept truly entails. They are selling a vision that often feels predestined, yet it is crucial to remember that this future is not fixed. We have engineered a technology capable of human-like feats, and this remarkable capability often obscures the fundamental truth that it is, at its core, still just a technology.
Undeniably, something monumental is unfolding—perhaps even comparable in scale to the invention of electricity or the internet. However, technologies of such transformative power invariably take time to fully integrate, mature, and bring about their lasting, profound changes. The full impact of AI is a story still very much in the making.
Source: MIT Tech Review – AI
