In March 2024, Kamil Jarzombek published a video analyzing a controversial statement by Sundar Pichai, CEO of Google. "It is possible that we will go extinct because of artificial intelligence" – these words still stir emotions. Two years later, the question remains relevant: does AI truly threaten us, or is it just another technological panic?
In March 2024, a video titled Google CEO: "it is possible that we will go extinct." Does AI really threaten us? appeared on the Kamil Jarzombek channel. Although more than two years have passed since its release, the topic still sparks heated discussions – especially in the context of the dynamic development of artificial intelligence in 2026. Are the warnings from tech leaders like Sundar Pichai justified, or are we dealing with an overreaction to a new technology?
Where did Sundar Pichai's statement come from?
In May 2023, Sundar Pichai, CEO of Google and Alphabet, gave an interview to The New York Times, in which he addressed potential risks associated with the development of artificial intelligence. His words were unambiguous: "It is possible that we will go extinct because of artificial intelligence. It is not excluded." Pichai emphasized that while we do not fully understand the mechanisms of human intelligence, we are trying to create systems that mimic it – and this raises fundamental questions about control over such technologies.
Pichai's statement was not isolated. Other industry leaders spoke in a similar tone, including Geoffrey Hinton, known as the "godfather of neural networks," who left Google in 2023 to speak freely about the risks associated with AI. Hinton warned that artificial intelligence could become "more intelligent than humans" and spiral out of control. Meanwhile, Elon Musk has repeatedly emphasized that AI represents "one of the biggest threats to civilization."
Are these fears justified, or are we dealing with excessive drama? To answer this question, it is worth looking at the arguments for and against.
Arguments for: Why could AI be a threat?
1. Expert statements and reports
One of the most frequently cited arguments is the fact that warnings about the dangers of AI do not come from random people, but from recognized scientists and tech leaders. Geoffrey Hinton, who worked on the development of neural networks for years, stated in 2023 that "AI could become more intelligent than humans within the next few decades." His concerns primarily relate to the possibility that AI systems might start acting in ways inconsistent with their creators' intentions.
A similar position is presented by the Asilomar AI Principles from 2017, signed by, among others, Stephen Hawking and Elon Musk. The document contains 23 ethical principles for AI development, including a warning about the risk of losing control over superintelligence. Furthermore, the 2022 report Existential Risk from Power-Seeking AI, prepared by the University of Oxford, estimates that the risk of human extinction by AI within the next 100 years is between 5 and 20 percent.
2. Threat scenarios
Experts point to several potential scenarios in which AI could threaten humanity:
- Misalignment problem: AI may pursue goals inconsistent with the creators' intentions. A classic example is the so-called paperclip maximizer – a hypothetical system whose goal is to maximize paperclip production. In an extreme scenario, it could consume all of Earth's resources to achieve its goal, disregarding the consequences for humans.
- Autonomous military systems: AI-based weapons, such as drones or cybernetic systems, could make decisions without human oversight. The 2017 Slaughterbots report warned against such a scenario, highlighting the risk of escalating armed conflicts.
- Loss of control over AGI: Artificial General Intelligence (AGI), i.e., systems capable of performing any intellectual task at a human level or higher, could act unpredictably. If AGI gained the ability to self-improve, it could quickly surpass human control capabilities.
3. Examples from the history of technology
History shows that new technologies often raise concerns that do not always materialize. However, some have indeed brought serious threats. An example is nuclear energy, which provides clean energy on one hand, but creates the risk of disasters (e.g., Chernobyl, Fukushima) or the proliferation of weapons of mass destruction on the other. Similarly, AI, while potentially bringing enormous benefits, carries the risk of uncontrolled development.
Arguments against: Why might the fears be exaggerated?
1. AI has no consciousness or intentions
Many experts emphasize that current artificial intelligence models, such as Large Language Models (LLMs), do not possess consciousness or the ability to think independently. Yann LeCun, Chief AI Scientist at Meta, stated in 2023 that fears of a "machine rebellion" are "science fiction." According to him, AI has no self-awareness and operates solely based on the data it was trained on.
Gary Marcus, a professor of psychology and neuroscience at NYU, compares current AI models to "stochastic parrots" – systems that generate responses based on patterns in data but do not understand their meaning or have the ability to plan long-term. This means that while AI can make mistakes or generate harmful content, it is incapable of acting with premeditation.
2. Regulations and control
In recent years, a series of actions have been taken to regulate AI development and minimize risks. An example is the EU AI Act, adopted in March 2024. It is the world's first comprehensive law regulating artificial intelligence, which introduces, among other things, a ban on high-risk systems such as emotion recognition in public spaces or social scoring.
Furthermore, in 2024, Google, Microsoft, OpenAI, and Anthropic established the Frontier Model Forum, an initiative aimed at promoting the safe development of advanced AI models. The companies committed to adhering to principles of responsible technology development and to cooperating with regulatory bodies.
3. The benefits of AI outweigh the risks
While the discussion about AI-related threats is important, one cannot forget the enormous benefits this technology brings. In 2026, AI is used in almost every area of life:
- Medicine: AI models help in diagnosing diseases, developing drugs, and personalizing therapies. An example is AlphaFold 3 from DeepMind, which revolutionized protein structure prediction.
- Climate: AI supports the fight against climate change, e.g., by optimizing energy consumption or forecasting extreme weather events. You can read more about this in the post How is artificial intelligence revolutionizing wind energy in 2026?.
- Science: AI accelerates scientific research, e.g., in astronomy or biology. An example is the Galaxy Zoo project, where AI helps classify galaxies based on telescope images.
Why do tech leaders talk about the threat?
Statements by tech leaders regarding AI-related threats may have various motivations. Some, like Geoffrey Hinton, left tech companies to be able to speak freely about the risks. Others, like Elon Musk, may use the topic to promote their own projects (e.g., Neuralink or xAI).
There is also the aspect of industry competition. Google and OpenAI are competing for market dominance in AI, and highlighting risks can serve to slow down competitors by introducing restrictive regulations. On the other hand, tech companies may fear regulatory pressure and try to get ahead of it by declaring a responsible approach to AI development.
What can we do to minimize the risk?
Kamil Jarzombek's video not only analyzes the threats but also presents proposals for actions that can help minimize the risks associated with AI. Here are a few of them:
1. Conscious use of AI
At an individual level, conscious use of AI-based tools is key. This means, among other things:
- Verifying information generated by AI models, especially in contexts requiring high accuracy (e.g., medicine, law).
- Avoiding reliance solely on AI when making important decisions.
- Supporting organizations involved in AI ethics, such as the Future of Life Institute or the AI Now Institute.
2. Responsible AI development in companies
Tech companies should implement Responsible AI principles, which include:
- Transparency regarding how models work and the data used to train them.
- Making models available for external audits, e.g., through Bug Bounty programs for AI.
- Investing in AI Safety research, i.e., techniques ensuring that AI systems will act in accordance with human values.
3. Regulations and international cooperation
At the level of countries and international organizations, the following are key:
- Cooperation on global regulations, such as the UN Global Digital Compact, negotiated since 2023.
- Funding research on alignment – adjusting AI goals to human values.
- Developing explainable AI techniques that allow us to understand how models make decisions.
4. Education and social dialogue
It is important that the topic of AI is widely discussed not only in scientific circles but also in society. Education about the capabilities and limitations of AI can help avoid both excessive optimism and unjustified fear. An example of such a dialogue is Kamil Jarzombek's video, which attempts to present the topic in a balanced way accessible to a wide audience.
Updates: What has changed since the video's release?
Since the video's release in March 2024, much has changed. Here are the most important events that took place up to July 2026:
1. Technological progress
- Google Gemini 1.5 (February 2024): A model with a 1M token context window, capable of analyzing long documents and videos.
- OpenAI GPT-5 (March 2025): The first model with declared capabilities for autonomous learning, albeit to a limited extent.
- Meta Llama 3 (April 2024): An open model competitive with GPT-4, available under an open-source license.
- AlphaFold 3 (2025): A DeepMind model predicting protein structures with accuracy close to laboratory experiments.
- Project Q (2026): An experimental Google project combining language models with physical robots.
2. New regulations
- EU AI Act: In August 2024, the first provisions came into force, including a ban on social scoring. In 2025, work began on the AI Liability Directive, regulating liability for damages caused by AI.
- USA: In 2025, the Senate considered the AI Foundation Model Transparency Act, requiring companies to disclose data about model training.
- UN: In March 2026, the General Assembly adopted a resolution calling for international cooperation regarding AI Safety.
3. The debate on risks in 2026
- In 2025, Demis Hassabis, CEO of DeepMind, stated that "AGI could emerge within a decade" and called for a global moratorium on research into the militarization of AI.
- In April 2026, a group of scientists, including Stuart Russell, published an open letter calling for a halt to work on models with parameters above 100 trillion.
- In June 2026, unconfirmed information leaked about Google's AutoCode project, which tested autonomous AI agents capable of executing code independently.
Summary: Does AI really threaten us?
Kamil Jarzombek's video does not give a definitive answer to this question, but it presents a balanced analysis of the topic. Here are the most important conclusions:
- Risk exists, but it is uncertain: Tech leaders take the threat seriously, although there is no evidence that it is inevitable. The concerns primarily relate to future AGI systems, not current models.
- Responsibility is needed: Regulations, transparency, and AI Safety research are key. Without these actions, the risk may increase.
- Do not panic: History shows that new technologies often raise fears that do not always materialize. AI can bring more benefits than risks if developed responsibly.
In 2026, artificial intelligence is already an inseparable part of our lives. The question is not whether AI will develop, but how it will do so – and whether we will be prepared for it. That is why it is so important that the discussion on this topic is conducted in a reliable manner, based on facts, and taking into account both the risks and the opportunities that this technology brings.
If you are interested in how AI is changing other areas of life, we invite you to read our previous posts, e.g., Has artificial intelligence become a "second intelligence"? or How is artificial intelligence changing science in 2026?.
Comments