Artificial intelligence is no longer simply improving our operate – it’s improving itself. This is radically accelerating the pace of development and could lead to an intelligence explosion.
February 22, 2026, 10:24 PMFebruary 22, 2026, 10:24 PM
The rapid advancements in AI are prompting a reassessment of its potential impact on the global economy and workforce.
Since ChatGPT launched in November 2022, artificial intelligence (AI) has catapulted into public consciousness and become established in the daily routines of many office workers. From answering emails and writing reports to creating illustrations and tables, chatbots like ChatGPT, Gemini, and Claude are being utilized for a wide range of tasks. Some view it as a mere novelty, others as a genuine efficiency gain, and still others as an indispensable tool.
When artificial intelligence becomes a model student in the workplace.image: getty images
However, AI has yet to trigger a true revolution in the workplace and beyond. Whereas some industries have experienced layoffs directly attributable to AI development, widespread job losses haven’t materialized, and entire sectors haven’t collapsed – as frequently predicted.
Is AI therefore not as revolutionary as some believe? Curious outputs from ChatGPT still draw amusement, revealing the limitations of its “brain” wired with hundreds of thousands of chips. Meanwhile, observers are watching Wall Street to see if the AI bubble in the stock market will burst.
Is it all just hype? Two recent developments suggest otherwise, indicating that we may not be able to afford to feel too secure regarding machine intelligence. The pace of change could be faster than anticipated.
AI developer Matt Shumar recently compared the current situation in an essay that gained traction on X to the period just before the COVID-19 pandemic in early 2020. Virologists warned that something significant was coming, but few listened or could imagine how drastically life would change in the following months.
What are these two developments?
First: AI Agents Can Do Much More Than Chatbots
While many people simply chat with ChatGPT, often using free versions that are still based on the November 2022 model, they overlook the broader capabilities of AI. It’s no longer limited to a small chat input field, but can independently perform actions on a computer or the internet, such as reorganizing photos or files.
An agent “sees” the screen, moves a virtual mouse, clicks through menus, opens a spreadsheet, copies data from emails, and enters it into a form. Or it can independently book hotels online. Granting access to a credit card or Bitcoin wallet allows it to handle payments or trading. Virtually anything that can be executed on a screen can be taken over by AI.
Companies have been cautious in publicizing this functionality, and AI agents are currently primarily used by tech specialists, due to the potential for errors that could have serious consequences if given access to all files or even credit card information. This hasn’t stopped companies from continuing to develop and improve the technology.
The situation is similar to the launch of ChatGPT: while established tech giant Google hesitated to launch its AI chatbot, newcomer OpenAI took the lead. It’s likely that a company will soon release its AI agents to the general public.
In software development, AI agents are already indispensable to experts, which leads to the second development that will massively accelerate the progress of AI.
Second: AI is Programming Itself
When OpenAI developed the first version of ChatGPT, it took over a year to produce a new version of the language model.
In 2025, multiple updates were released per year – now they are being developed almost monthly. Recently, both OpenAI with GPT-5.3-Codex and Anthropic with Claude Opus 4.6 simultaneously unveiled new, powerful models. This acceleration in intelligence development is due to programmers using AI to develop even better AI.
Dario Amodei, CEO of Anthropic, says that AI now writes “a large part of the code” in his company. According to the company, the new ChatGPT model is the first to have been significantly involved in its own development. The recently released version 5.3 is being used to develop an even better version 5.4 – and so on.
Dario Amodei, CEO & Co-Founder of Anthropic.image: AP
leaps in capability are not only happening faster, they are also becoming larger, leading to an explosion of intelligence. Amodei believes that we are “only one to two years away from the current generation of AI autonomously developing the next one.”
This doesn’t necessarily lead to superintelligence or consciousness in machines – although Amodei doesn’t rule it out. But it has significant consequences for us.
Amodei illustrates this with a thought experiment: imagine it is 2027. Overnight, a new country with 50 million citizens appears, and each individual is smarter than any Nobel laureate who has ever lived.
You wouldn’t want to antagonize such a state, but rather cooperate with it. For the world political stage, this means that it is crucial that the “new people” belong to the “quality” ones, meaning that authoritarian regimes like China do not win the AI race. And for all of us in the workplace?
The CEO of Anthropic’s prediction: AI will eliminate 50 percent of entry-level office jobs within one to five years. But as the appear at his own company shows, even highly qualified jobs in the IT industry are not safe.
we should be interested in working with these AI Nobel laureates, having them on our team, and leading them.
Matt Shumar’s advice, comparing the situation to the false sense of security before the coronavirus pandemic, was to adapt as quickly and as well as possible. Most jobs will not simply be replaced by AI, but by a person who knows how to use AI best.
