Google DeepMind Researcher Says AI Could Pass Human Level
Google DeepMind research scientist Tim Genewein says artificial intelligence could progress beyond human-level performance across a broad range of tasks. He argues that researchers and policymakers should prepare for the possibility, while acknowledging that limits on AI capabilities could stand in the way.

Google DeepMind research scientist Tim Genewein says artificial intelligence could pass a major threshold: matching human performance across nearly all cognitive tasks. He argues that AI could then progress to superhuman capabilities, making preparation a challenge for researchers, developers and policymakers.
Artificial general intelligence, or AGI, describes proposed machines that can match human performance across nearly all cognitive tasks. Artificial general superintelligence, or ASI, would go further, outperforming large numbers of people across a wide variety of tasks.
Genewein says superintelligent systems could make scientific progress across an entire research field in a short time. Chip design is another example of a task such systems might handle.
To qualify as ASI, computers would need to reach that standard across nearly all tasks of interest to humans. Superior performance in a few specific areas would not be enough.
Computing power drives forecast
Genewein says improvements in chips and AI efficiency, combined with investment in hardware, are increasing effective computing power for AI work tenfold each year. He describes the process as “cognitive industrialisation.”
He says those gains amount to an improvement of 10 billion times over ten years. Even without dramatic advances in AI itself, Genewein argues, more resources for existing systems could bring transformations.
Greater computing power might help bring human-level AI within reach. It would not necessarily overcome limits that could prevent a further leap to superintelligence, Genewein cautions.
He compares the potential limits with Albert Einstein's discovery of general relativity. Even a thousand astronomers from the ancient world working together would not have made the same discovery, Genewein argues.
Limits and preparation
Some critics question whether AGI is possible at all. They argue that current AI systems face fundamental barriers, including their reliance on language and limits on what they can know about the physical world.
Genewein points to several developments that could change the picture. They include new training data, such as models of the real world, and technology that could help systems invent ideas rather than rework old ones.
AI agents might also work together more effectively. Another possibility is recursive self-improvement, in which AI systems train themselves and potentially accelerate their own progress.
Genewein says preparation will require efforts across leading AI labs and national and global institutions. He also calls for new ways to measure how quickly AI capabilities are improving.
“Human-level AGI has long been the horizon for AI research, but it is far from the upper limit of intelligence,” Genewein writes.
Take the quizWhat Kind of Holiday Spender Are You? Take the Black Friday QuizStart →- 1Stock Market Today: Dow Slides 1.34% and VIX Jumps 15% at Midday

- 2Joachim Klement to Sell Stocks After S&P 500 Signal

- 3Kenneth Walker’s Groin Injury Puts Cardinals Game in Doubt

- 4Pat Cummins Faces Captaincy Doubts From Mitchell Johnson

- 5Stock Market Today: Nasdaq, S&P 500 Futures Slide as AI Selloff Deepens

Lifestyle · Food & HealthNormal Blood Pressure Chart by Age: Healthy Ranges Explained
Lifestyle · Food & HealthWhat Is Raynaud’s Syndrome? Why Fingers Turn White in the Cold
Lifestyle · Food & Health4 Ways Walking After Meals Helps Your Body, Doctors Say
Lifestyle · Food & Health4 Reasons to Eat Oatmeal, According to Nutrition Experts














