I expect AI to First excel at any kind of online work - essentially anything that a remote worker can do. AI will do better at tasks like copywriting, tax preparation, customer service, and many other jobs. These are areas that involve reading, analyzing, and synthesizing information, and then generating content based on it. These tasks seem ripe for replacement by language models.

The pace of improvement in robotics lags significantly behind cognitive automation. Jobs that involve intricate, guided, specific physical work - like surgeons, gardeners, plumbers, jewelry makers, and hair stylists - may find their skills contributing to society for many more years to come. Regulated industries like medicine and the civil service will also have longer human involvement, though even there I expect an increasing number of human workers to be supplemented by AI systems.

For the rest of the economy, I expect AI to eventually be able to do every economically useful task. Many expect AI to reach this level of capability. Given the current trajectory of technology, I believe AI will first excel at any kind of online, remote, or knowledge work. These are the areas that seem most vulnerable to automation in the coming years.

Human workers in these domains may find themselves less economically valuable than AI systems that can work 24/7 without breaks. Companies will have strong incentives to replace high-salary human workers with lower-cost AI alternatives. Individuals in these fields need to think carefully about how to position themselves for this transition.

Human skills that are more difficult to automate, like empathy, creativity, and physical dexterity, may become more highly valued. But the overall trend seems to be towards the obsolescence of a large swath of knowledge work as AI systems rapidly improve.

According to the transcript, there are certain careers that may be preferred for humans even as AI systems become more advanced. These include:

• Counselors
• Therapists
• Caretakers for the elderly
• Babysitters
• Preschool teachers
• Priests and religious leaders
• Sex workers

The key point is that these roles often involve a human-to-human connection and empathy that AI systems may struggle to replicate, at least in the near future. Even as AI becomes more capable, there may be a "nostalgic" preference for human workers in these types of roles.

The transcript notes that "much has been made of AI girlfriends, but I still expect that a large percentage of buyers of intimacy and sexual services will have a strong preference for humans." This suggests that certain services requiring a personal, emotional connection may remain dominated by human workers.

Overall, the careers listed seem to be those that rely heavily on subjective human experiences, empathy and interpersonal skills - areas where AI may have difficulty matching or surpassing human capabilities in the immediate future, despite rapid advancements in language models and other AI technologies.

Human workers in these fields may find their roles more secure in the coming years of AI development and automation, though the long-term trajectory remains uncertain.

Evidence shows, unsurprisingly, that unemployment does appear to increase the likelihood of reporting poorer health and well-being. One study that tried to disentangle the causality by looking at the effects of unemployment caused by the collapse of the Spanish construction industry found that unemployment did increase the likelihood of reporting poorer mental and physical health.

However, it isn't entirely clear if this is an inherent fact of unemployment or a contingent outcome. The psychological effects may depend on factors like whether the unemployment feels like a personal failing versus a blameless, shared experience. For example, the study found that individuals who were temporarily laid off in April 2020 reported lower levels of distress compared to their peers who remained employed. The widespread nature of the layoffs normalized the experience, reducing personal blame and fostering a sense of shared experience. Financial strain was also mitigated by government support, personal savings, and reduced spending.

This suggests the psychological effects of losing work in a post-AGI world may depend on how the transition is framed and experienced. If it is seen as a blameless, shared experience with a social safety net, the negative impacts may be less severe than if it feels like a personal failing.

At the same time, there may be a deeper psychological need to feel useful and contributing that abundance alone cannot satisfy. An AI researcher mentioned practicing activities he's not good at, like jiu-jitsu and surfing, to prepare for a future where we may have to find meaning and joy in activities rather than need or excellence. Navigating this transition, both economically and psychologically, will be a major challenge.

The future of employment is rapidly changing, and it's crucial to understand the implications of the advancements in artificial intelligence (AI). As Avital Balwit, the Chief of Staff at Anthropic, eloquently states, the next few years might be the last few years that she works. This is a profound statement that highlights the gravity of the situation we are facing.

Balwit's insights into the impact of AI on the job market are invaluable. She notes that with each iteration of Anthropic's AI models, she is confronted with something more capable and more general than before. This rapid progress in language models and cognitive automation is poised to disrupt a wide range of industries, including copywriting, tax preparation, customer service, and even software development and contract law.

The key point Balwit makes is that the economically and politically relevant comparison is not whether the AI system is better than the best human, but whether it is better than the human that would otherwise do the task. This distinction is crucial, as it shows that automation will occur in any scenario where the AI system is more efficient than the average human worker.

Balwit also highlights the fact that the pace of improvement in robotics lags significantly behind cognitive automation. This means that jobs involving manual labor, such as those in the medical, construction, and creative industries, may be less affected in the short term. However, she expects that even in these regulated industries, human workers will increasingly be supplemented by AI systems.

The article also delves into the psychological and social implications of a post-AGI world. Balwit acknowledges the potential negative effects of widespread unemployment, such as increased levels of distress and anxiety. However, she suggests that these effects may be mitigated if the displacement is widespread, as it can foster a sense of shared experience and reduce the personal blame associated with job loss.

Balwit's insights into the concept of "nostalgic jobs" are particularly intriguing. She suggests that certain roles, such as counselors, caregivers, and religious leaders, may be less susceptible to automation due to the inherent human connection and empathy required. These "nostalgic jobs" may become increasingly valuable in a world where AI systems excel at cognitive tasks.

Finally, Balwit discusses the importance of finding meaning and purpose beyond traditional employment. She suggests that individuals should prepare for a future where they may need to engage in activities not for the sake of excellence, but for the joy of the experience itself. This mindset shift could be crucial in navigating the challenges of a post-AGI world.

In conclusion, Balwit's insights provide a thought-provoking and sobering look at the future of employment. It is clear that the advancements in AI will have a profound impact on the job market, and it is essential for individuals to be proactive in preparing for this transition. By understanding the trends and potential implications, we can better position ourselves to thrive in the post-AGI world.

The key points from this insightful article by Avital Balwit, the Chief of Staff at Anthropic, are:

1. Impending Obsolescence of Knowledge Work: Balwit expects AI systems to soon excel at a wide range of knowledge work tasks, from copywriting to software development. This poses a significant threat to many white-collar professions.

2. Uneven Automation Pace: The pace of automation will vary across industries. Manual, physical jobs may be more resilient than cognitive, desk-based work. Regulated fields like medicine could see slower AI adoption.

3. Psychological Impact of Joblessness: The psychological effects of widespread joblessness are unclear. Universal basic income and the "normalization" of unemployment could mitigate distress. However, the loss of a sense of purpose and contribution may still be challenging.

4. Preparing for a Post-Work Future: Balwit suggests practicing activities one is not particularly skilled at, finding joy in the process rather than the outcome. This may help adapt to a future where one's economic value is not the primary source of meaning.

5. The Role of Aligned AGI: Balwit speculates that advanced, benevolent AI systems may be able to help address the problems they create by assisting humans in finding purpose and fulfillment without traditional employment.

This article provides a sobering, yet thoughtful, perspective on the impending changes to the nature of work. It highlights the importance of proactively preparing for this transition, both individually and as a society. The insights shared here deserve serious consideration as we navigate the coming technological revolution.