Here’s another example of a different way to think about language models: ask it for concrete advice when you’re faced with a difficult decision. In particular, the kinds of decisions that need to integrate lots of context, many variables, with high levels of uncertainty, and high-stakes outcomes. The kinds of decisions that, even with the best of intentions, are little more than guesses.
Here are a few common causes of bad decisions:
- Assumptions based on small sample size.
- Wanting the world to work the way we want rather than the way it does.
- Conforming to expectations/authority/group (social default).
- Blindness to large trends (blind spots).
- Not asking, “and then what?”
These are all scenarios that a conversation with Claude can help avoid.
Barack Obama suggested three strategies for better outcomes in high-stakes decision-making:
- Assign probabilities to a range of outcomes.
- Seek out information from people who are smarter than you
- Ask more questions: you don’t have to have all the answers.
Again, these are all things you can ask Claude to do for you.
Here is an example prompt for someone considering a career change in a context where the decision isn’t simple (I think you could probably get a decent response without giving this much context): “I need to make a decision about changing careers. I really enjoy the work I currently do, it pays relatively well, I feel like I have good work-life balance, I get to work flexible hours, and I get along well with my co-workers. So, I have no ‘push’ factors that are forcing me to consider other options. However, I’ve recently noticed other opportunities in the same sector, but with a different emphasis in the role I’d be playing. There isn’t much difference in pay, I’d have the same flexibility, and the general focus of my attention would be the same. However, the ‘pull’ factors for this other job are significant, making me wonder if the challenge of a new direction, and the potential for making a different kind of impact, are making me wonder if I should explore other career options. What would you do, if you were in my position? How should I evaluate the possible outcomes of different decisions? What kinds of questions should I ask, and who should I be asking? What other information can I gather that might help me make this decision? What if I make the wrong decision and it’s not reversible?”
Response from Claude
Here are a few suggestions on how to evaluate your career options:
- Make a list of pros and cons for staying in your current role versus changing to the new opportunity. Consider things like work environment, opportunities for growth, compensation, work-life balance, etc.
- Reflect on your values and priorities. What’s most important to you in a job? Is it the type of work, the impact you can make, work culture, compensation, etc.? Knowing what matters most can help guide your decision.
- Discuss with people you trust – mentors, colleagues, friends and family. Get their input on what they think would be the best fit for you. They may provide perspectives you haven’t considered.
- Ask about growth opportunities in each role. Will one option provide more chances for advancement in the long-run?
- Consider trying the new opportunity as a temporary assignment or side project first, if possible. This can give you direct experience before fully committing.
- Trust your instincts. Reflect on how each option makes you feel when you imagine yourself in that role. Our intuition often guides us in the right direction.
- Remember that very few decisions are completely irreversible. Even if you try the new role and later realize it’s not the right fit, you can likely transition back or find another opportunity.
The key is gathering as much information as you can, weighing the pros and cons, and reflecting on your career goals and values. An informed decision is the best one. Changing careers is challenging but can reap rewards if it’s the right move for you.
That’s not a bad starting point to begin reflecting on a difficult decision like changing careers.
Obligatory caveat: I’m not suggesting that you blindly do what Claude tells you.