The article you shared raised many important points about the limitations of AI and the enduring value of human skills. To provide a comprehensive response to the piece, I have drawn upon the findings from the 2023 AI Index report from Stanford University's Human-Centered Artificial Intelligence (HAI) initiative and other resources.
The report highlights several key points that align with the arguments presented in your article. First, large language models (LLMs) such as GPT-3 and PaLM have indeed scaled up significantly in size and expense, which shows how much reliance and investment there is in AI technologies. However, these models, despite their complexity and sophistication, cannot inherently possess human-like understanding or judgement. They cannot resolve conflicts or apply critical thinking in the same way a human can. Their responses are based on patterns learned from training data and lack the nuanced understanding that comes from lived human experience.
The report also points out that current AI tools continue to meet or exceed technical benchmarks, indicating increased capability. Yet, they still require more challenging tasks to be fully tested, underscoring the ongoing necessity for human oversight and intervention.
The environmental costs of training these models are also a significant concern. For example, GPT-3 was identified as a heavy carbon emitter, and even more efficient models still consume substantial amounts of energy. This underlines the fact that AI, despite its advantages, comes with substantial trade-offs that need to be carefully managed.
The report also mentions a surge in AI controversies, with reported issues being 26 times greater in 2021 than in 2012. This speaks to your point about AI's limitations in conflict resolution and the potential for misuse.
Despite these issues, the demand for AI labor is growing, and the number of AI job postings was notably higher in 2022 than the previous year. This suggests that rather than making humans obsolete, AI is creating new job opportunities, albeit in different sectors and roles.
Interestingly, the report also mentions a significant increase in AI-related legislation passed in different countries. This indicates a growing awareness and proactive response to the ethical, societal, and economic implications of AI.
