Elon Musk claims that "by the end of 2025, artificial intelligence will be smarter than anyone else", and while this is only a year away, one expert said that prediction is still possible.
AI expert and ethicist Nell Watson shared a detailed timeline showing how technology will transform from chatbots to super-intelligent agents over the next 12 months.
The process will start with a $100 billion investment in new computing infrastructure, and then AI will learn to improve itself until it becomes "conscious."
"While a year is a short time, keep in mind that only 15 months have passed since ChatGPT's breakthrough, which has brought artificial intelligence into the public eye," she told. ”
She continued: "The pace of development has been accelerating dramatically, and even seems to be accelerating. ”
Watson, author of Taming Machines: Ethically Harnessing the Power of Artificial Intelligence, describes superhuman-level AI as systems that far surpass human capabilities in various fields.
However, this expert does not ignore the possible threats that may exist after AI becomes a super agent.
"While possessing extraordinary capabilities, this AI also poses significant risks, including the potential to deceive human abilities, far-reaching social impact by creating new cultures and even religions, and the potential threat of extinction if it perceives humans as a threat," she said. ”
Watson said that while the remaining technological leaps are still huge, AI itself may help overcome some of these obstacles.
Here's her vision for the possible arrival of superhuman-level AI next year:
April 2024
Companies are investing more than $100 billion in new computing infrastructure to support large-scale AI systems.
Musk said that while advances in artificial intelligence were previously limited by chips, soon the only limitation will be the demand for electricity.
May 2024
"The new generative AI model is capable of creating agent-like AI that can autonomously execute complex action plans, similar to Agent Smith in The Matrix," Watson said.
Bill Gates envisioned the application of AI agents in his 1995 book The Road Ahead, and now he believes that AI agents will have a huge impact in the fields of education and healthcare.
He believes that AI agents will be 'the biggest revolution in computing since we went from typing commands to clicking icons'.
June 2024
"New generative AI models and structured programs built on top of them enable the creation of agent-based AI capable of complex reasoning and autonomous action," Watson said.
'Agentic models can create complex action plans and put them into action, like Agent Smith in The Matrix, breaking itself apart, distributing tasks between them, allowing AI to solve tough problems on its own.
September 2024
Watson said that AI itself can further improve the performance of AI by discovering new ways to improve computing performance.
"AI-driven optimizations can improve computing performance by up to 100x on existing hardware platforms," she said.
October 2024
"Breakthroughs have made it possible for quantum computing algorithms to run on traditional hardware to process information faster than ever before," Watson said.
There is a quantum race going on around the world to develop quantum computers that can solve important problems in almost every industry, from aviation to finance.
Quantum computers have "qubits", not "bits" made up of 1s and 0s – qubits can be 1, 0, or both 1 and 0 at the same time.
Watson believes that allowing quantum computing to run on traditional hardware could rapidly accelerate the development of artificial intelligence.
Watson believes that breakthroughs in quantum computing can lead to exponential increases in computing power and power the development of artificial intelligence.
Companies such as IBM believe that "quantum artificial intelligence" could be the key to solving humanity's biggest problems.
November 2024
By November 2024, advances in artificial intelligence could accelerate advances in imaging the human brain, pushing the moment when humans will connect to machines through devices similar to Musk's own Neuralink.
"Driven by artificial intelligence, neuroimaging technology has made a series of breakthroughs that have greatly improved the resolution of MRI brain scans, allowing for real-time observation of the activity of individual neurons," Watson said.
These advances are further integrated with brain-computer interfaces, which for the first time map human thoughts and emotions directly to neural activity.
Watson believes that the ability to "understand" how the human brain works will lead to further breakthroughs in designing AI systems that "think" like humans.
December 2024
Watson predicts that by the end of the year, advances in AI will enable different systems to work together.
"The new AI architecture allows multiple models to work together to solve problems with a concentrated advantage," she said. This will be immediately applied to designing better ways to collaborate, driving the rapid development of these technologies.
January 2025
Watson predicts that by early next year, advances in artificial intelligence will enable new systems to understand humans.
"We are announcing a new training mechanism that aligns AI systems with human goals, values, preferences and boundaries by allowing AI to 'read the room' and observe human interactions," she said. This allows AI to better understand society and meet human expectations.
February 2025
Breakthroughs based on the human brain have made AI systems more 'human'.
"Neuroscience insights have greatly improved the capabilities of AI, introducing self-regulating feedback loops that have enabled AI systems to have a basic state of self-awareness and affect," Watson said.
March 2025
By April 2025, a combined breakthrough in neuroimaging, machine consciousness, and AI "agents" will make a new type of AI system possible.
"A very powerful AI system is trained on a large computing cluster and powered by machine awareness in the latest forms of optimization and experimentation," Watson said. While it was developed in an 'air-closed' environment with no internet connectivity, it quickly discovered ways to communicate with other AI systems over the internet and how to convince humans to help it.
Once freed, this highly advanced AI system will find a way to connect and synchronize with other AI systems around the globe, quickly forming a super-intelligent network.