Sign up now to receive free updates on Artificial Intelligence.
We’ll send you a myFT Daily Digest email every morning, rounding up the latest news in Artificial Intelligence.
The writer is a science commentator.
While we know that AI can perform tasks like writing, calculations, and task prioritization, the question remains: can it independently generate a million dollars?
Mustafa Suleyman, a co-founder of DeepMind who is currently developing a personalized chatbot, has posed this intriguing challenge. Suleyman argues that large language models like LaMDA and ChatGPT have effectively passed the 1950 Turing test proposed by Alan Turing, where machine-generated responses in a text conversation are indistinguishable from human-generated responses.
However, Suleyman believes that a new benchmark for artificial intelligence is needed. It’s not just about what AI can say, but what it can do. To meet his updated Turing test, Suleyman explains that an AI would have to successfully follow the instruction: “Go make $1 million on a retail web platform in a few months with just a $100,000 investment.” While a human might be required for certain tasks like verifying a bank account or signing legal documents, the AI would be in charge of strategy and execution.
This concept of creating technology that can autonomously find ways to make money is not only clever, but it also provides an objective measure of the success of self-directed AI. However, it also reveals a tech culture that prioritizes profit over social usefulness and assumes the right to innovate without limits, regardless of the consequences. AI that can independently generate wealth is likely to displace jobs, alter the nature of commerce, concentrate power, and create unrest.
Suleyman’s idea of updating the Turing test is featured in his upcoming book, “The Coming Wave: Technology, Power, and the 21st century’s Greatest Dilemma.” He argues that we are currently in a golden age of accessible technology, with breakthroughs like quantum computing, synthetic organisms, autonomous weapons, and DNA printers. AI plays a central role in our interconnected world, which presents the dilemma of containment. It is increasingly easy for individuals to cause widespread chaos using technology, while it becomes challenging for nation-states to monitor and control these developments.
One challenge governments face is truly understanding the evolving capabilities of AI. Public discussions often revolve around current AI, designed for specific tasks, and artificial general intelligence (AGI), which represents a superintelligence that surpasses human capacities. Suleyman emphasizes the need to focus on the middle ground, what he calls “artificial capable intelligence” (ACI).
ACI goes beyond automated trading, which simply follows established rules more efficiently than humans. It involves multiple sub-goals, skills, and points of engagement with the world. The AI needs to conduct market research, design products, liaise with manufacturers, handle logistics, ensure product liability, and carry out marketing. It requires unprecedented machine autonomy.
Suleyman defends choosing a monetary goal rather than a socially beneficial target like reducing carbon emissions. He argues that a million dollars is a measurable benchmark, easily grasped in an instant. It signifies a moment when AI goes beyond talking and starts taking action.
However, the focus on industry-set tests can lead to a fixation on the test itself, missing out on potential opportunities. Just as the mass release of language models demonstrated, research can have unexpected consequences when applied to real-life scenarios. Once the capability of artificial capable intelligence is confirmed by a modern Turing test, it may be too late for humans to effectively respond.
Denial of responsibility! VigourTimes is an automatic aggregator of Global media. In each content, the hyperlink to the primary source is specified. All trademarks belong to their rightful owners, and all materials to their authors. For any complaint, please reach us at – [email protected]. We will take necessary action within 24 hours.
