👁 Humanity's Golden Age or its End? Where AI Will Take Us
In 2021, a year before the release of ChatGPT, former OpenAI Governance team employee Daniel Kokotajlo predicted that within five years, AI algorithms would learn to create realistic photos and videos, be able to diagnose more accurately than doctors, hit a scaling "wall," and AI agents will start its expansion.
In April 2024, he left OpenAI due to disagreements with the company's AI safety policies. Teaming up with a group of researchers, he has recently unveiled a new report with forecasts about AI development for the next five years. Its main characters are the fictional US company OpenBrain and its Chinese competitor DeepCent.
⌚️ Main predictions:
2025-2026: AI agents become more autonomous and master all programming languages. OpenBrain builds the largest data center in history, and DeepCent opens a computing cluster at one of its most powerful nuclear power plants.
2027: OpenBrain releases an AI agent on a new architecture, which allows it to continuously improve by learning "online." China secretly steals the model's data but lacks the capacity to run the system fully. Meanwhile, OpenBrain claims to have achieved AGI and releases a mini-version of the next agent for everyone.
Kokotajlo's team believes that by the end of 2027, AI systems will surpass humans in everything.
2028-2029: The arrival of superintelligence. AI researchers will achieve efficiency levels 40 times greater than humans, paving the way for groundbreaking advancements in fusion technology, quantum computing, and the development of highly effective pharmaceuticals.
The AI economy is growing rapidly: businesses and laboratories run entirely by artificial intelligence are emerging worldwide, and poverty is eliminated with universal basic income.
😎 Utopia or cyberpunk?
According to Kokotajlo and his team, if governments can overcome conflicts, agree, and ensure transparent use of superintelligence, then a technological breakthrough in the 2030s could boost the colonization of the Solar System.
The apocalyptic scenario includes the arrival of a reality similar to The Matrix movie, where artificial intelligence enslaves and potentially destroys humans and all living organisms.
Kokotajlo himself refers to himself as a techno-pessimist. He believes that AI will catastrophically harm humanity with a 70% probability.
What do you think: will AI lead us to prosperity or a catastrophe?
👍 — I'm a techno-optimist; AI is a breakthrough
😈 — I'm a techno-pessimist; AI is a threat
🤔 — it's hard to say definitively
#news #forecasts #OpenAI @hiaimediaen
