Artificial intelligence (AI) has become a dominant force in our world since the advent of ChatGPT and GPT-4. With new capabilities being announced almost every week, the debate surrounding AI has intensified. Will it create a better world, or will it lead to job loss and human impoverishment? The prospect of artificial general intelligence (AGI), where AI can outperform humans in all cognitive skills, raises concerns about control and the potential dangers, as seen in movies like “Terminator.”
“The Coming Wave: AI, Power and the 21st Century’s Greatest Dilemma,” written by Mustafa Suleyman, co-founder of DeepMind AI, offers invaluable insights into this complex subject. Suleyman’s background in AI gives his views on the future of humanity with AI significant credibility.
Suleyman highlights four key characteristics of AI. First, it is an asymmetrical technology, capable of generating vast amounts of text and potentially enabling operators to launch lethal drones with minimal resources. Second, AI has the potential for hyper-evolution, surpassing the growth rate seen in any other field. Third, AI is not a narrow-use technology like an X-ray; it has the potential to permeate every aspect of daily life, society, and the economy. Lastly, AI possesses autonomy, making it self-learning and increasingly unpredictable.
However, the challenges of regulating AI are considerable. Unlike quantifiable problems such as regulating auto emissions, a general-purpose omni-use technology like AI is much harder to contain. Moreover, researchers and inventors, driven by curiosity and ambition, may be unwilling to limit their potential for innovation. The economic incentives for AI development are enormous, and the profits that can be made are unimaginable. Additionally, while nuclear power plants are tightly regulated, the building of AI tools remains accessible to individuals, raising concerns of the potential misuse of powerful AI by rogue actors.
Suleyman proposes several solutions to address these challenges. Implementing a strong technical safety framework, similar to those governing nuclear power, could ensure the responsible development of AI. Tough audits for new AI tools and organized tests to uncover weaknesses or failure risks are also necessary. Governments can learn from events like India’s annual “red teaming,” where controlled attacks are simulated to identify vulnerabilities.
As AI continues to advance, it is crucial to strike a balance between harnessing its potential benefits and mitigating the risks. Regulation and responsible development are essential for ensuring that AI serves humankind and our planet.
The source of the article is from the blog toumai.es