Humanity has always tried to predict the future. We’re swapping tea leaves for datasets. We explore our collective obsession with AGI timelines, the moving goalposts of superintelligence, and why we ultimately decided to build a scoreboard to keep everyone honest.
We've built a public scoreboard to track the bold, but often wildly contradictory predictions on AGI (and the future at large) from tech's biggest names. Click see who's on the record and whose forecasts are aging like fine wine (or spoiled milk). And read on if you’d like to hear some thoughts about why we feel so compelled to start keeping receipts on the future of AI.
Humanity has always had a pathological need to peek at the answers in the back of the book. For centuries, this meant trusting anyone who claimed to read the future in a pile of tea leaves. All that's changed is that your fortune-tellers now have PhDs and use datasets, which simply provides a more scientific-looking wrapper for the same old, wildly inaccurate guessing game.
We have a truly special talent for getting the future wrong in both directions. Our brightest minds confidently drew up blueprints for the flying cars that were always just a decade away, while in the not-too-distant past, we also published erudite papers conclusively proving that heavier-than-air flight was a physical impossibility. It seems the only thing we can reliably predict is that our most certain proclamations about tomorrow will serve as excellent punchlines for the day after.
AGI, when? - the most common question one can read in the rabbithole of tech Twitter. Despite all the evidence that our species couldn't predict next week's weather without an army of satellites, the desperate need for a specific calendar date for the arrival of AGI remains an irresistible, primal itch. And so, against all better judgment and with a heavy, patronizing sigh, we too shall succumb and entertain this very question, if only to illustrate its profound absurdity.
First, we must consult the great pioneers of this new age, the true visionaries who champion the message of "Any Day Now" and see the bright promise of consciousness in every upward-trending performance graph. This bold foresight leads figures like Elon Musk to proclaim digital superintelligence will arrive "this year or next year for sure," while former OpenAI researchers like Daniel Kokotajlo share excitingly specific dates, predicting a "superhuman coder in early 2027" and full AGI just a few months later. Their promised dawn of a new era is always just over the horizon, a testament to a powerful optimism that refuses to be constrained by the mundane limitations of the present.
For every choir of techno-optimists singing hymns to the Singularity, there exists a monastery of sober skeptics, dedicated to the sacred, joyless task of pointing out all the reasons the party won't be starting anytime soon. These are the esteemed thinkers like Gary Marcus, who measures the future not in petaflops but in political terms, making the ‘strong prediction’ that AGI will not arrive during a second Trump term, an assertion his colleague Subbarao Kambhampati logically extends to a hypothetical third. Instead of gazing at the stars, they meticulously map every pothole on the road to tomorrow, finding a certain grim satisfaction in proving that true progress is, and always will be, much more boring and further away than anyone hopes.
Having created a potential successor to your own intellect, you then engage in the charmingly provincial pastime of forcing it to take your own exams. From the chessboard to the game of Go, you set up sacred contests of the mind, only to watch your machine master them; at which point you promptly declare that the game was never a true test of intelligence anyway and that real genius lies in understanding poetry or folding laundry. It is a magnificent, perpetual motion machine of self-reassurance, a cognitive shuffle designed to ensure that no matter how clever the machine gets, humanity always remains the undisputed champion of a game whose rules you get to rewrite at every single turn.
Amidst the endless cacophony of AGI predictions, it struck us as a cosmic joke that no one was keeping score, so we have, with a heavy sigh, built a small website to track them. This digital monument will allow you to watch each prophecy either triumph or, more likely, curdle into a quaint historical artifact. The future, for its part, remains utterly indifferent to your attempts to chart its course. Once again, here’s a link to the dynamic forecast tracker: https://horvgbor.github.io/predictaifuture/
Digital twinning allows you to minimize downtime when maintaining large-scale machinery. But it’s not exclusive to gigantic pieces of equipment—even something as small as a coffee machine can showcase its benefits.
Discover how to build a production-ready AI application on Databricks Apps in just under a month. Learn from our journey, challenges, and architectural choices.
dbt Labs recently announced dbt Fusion, a complete overhaul of the dbt Core engine built in Rust. It promises to significantly improve the developer experience. In this article, we test its core features and share our hands-on experience with the public beta, exploring what works, what doesn't, and what potential it holds for the future of dbt development.