Evaluating Alignment Jams

JavaScript isn't enabled in your browser, so this file can't be opened. Enable and reload.

Your feedback is important to us! Please take your time to complete this survey and help us improve the Alignment Jam hackathons. Your responses will by default only be used in summary statistics and we ask when we may quote you directly. We keep all responses private. In appreciation of your time, all respondents will be eligible for a free book (Superintelligence, Human-Compatible, or The Alignment Problem). For any questions about the survey, please contact operations@apartresearch.com.

There are 4 parts:

Impact: How did the hackathon(s) change your trajectory.
Improvement: How can we make the experience better.
Your project: What have you done since submitting your project.
Demographics: A few questions about who you are.

Please read: The purpose of this survey is to understand how the hackathons have helped your path towards AI safety. We are very interested in understanding how the hackathons changed your mind, enabled you to do something you would otherwise not have been able to, or changed your personal impact and trajectory. The survey will take approximately 10-25 minutes and we invite you to engage with the questions deeply.

Please only respond if you have participated in a hackathon, even if you did not submit a project.

Which hackathons have you participated in?

Select all that apply.

AJ1: Language Model Hackathon (Sep 30, 2022)

AJ2: Interpretability Hackathon 1.0 (Nov 11, 2022)

AJ3: AI Testing Hackathon (Dec 16, 2022)

AJ4: Mechanistic Interpretability Hackathon (Jan 20, 2023)

AJ5: ScaleOversight Hackathon (Feb 10, 2023)

EAG Bay Area Thinkathon (Mar 1, 2023)

AJ6: AI Governance Ideathon (Mar 24, 2023)

AJ7: Interpretability Hackathon 2.0 (Apr 14, 2023)

AJ8: Verifiable Safety Hackathon (May 26, 2023)

ARENA Interpretability Hackathon (Jun 9, 2023)

AJ9: Safety Benchmarks Hackathon (Jun 30, 2023)

AJ10: Interpretability Hackathon 3.0 (Jul 14, 2023)

AJ11: Evaluations Hackathon (Aug 18, 2023)

AJ12: Agency Hackathon (Sep 9, 2023)

Any distillation hacakthon

Other:

Required

If you participated in multiple hackathons, please select the one that had the most influence on you.

When following questions are about a "hackathon", please respond with this one in mind. Otherwise, a question will ask about the "hackathon(s)".

How have the hackathons helped you?

Can you share a story about how the hackathon(s) might have changed your life or career trajectory, particularly regarding AI safety? What may have been different if you hadn't participated?

This is an important question and the more concrete you can be, the better. The purpose is to understand how specifically the hackathon(s) changed your trajectory, e.g. "I had an insightful conversation that made me decide to work full-time on AI safety" or "I used my project in a job application and they mentioned it as a top reason for my selection".

We encourage you to spend extra time here, and if your time is limited, please prioritize this question. We ask if we can share your answer in the next question.

Can we share the answer above?

These can often be personal but it helps us a lot to be able to communicate the individual valuable cases of impact.

Yes, with my name

Yes, anonymized (we will edit the text if it contains any personal information)

Other:

Prior knowledge

How knowledgeable did you feel about AI safety before the hackathon?

Not at all

Very knowledgable

Post knowledge

How knowledgeable did you feel about AI safety after the hackathon?

Not at all

Very knowledgable

Prior likelihood of AI safety work

How likely were you to work in a full-time AI safety role before participating (includes non-technical roles).

100%

Post likelihood of AI safety work

How likely were you to work in a full-time AI safety role after participating (includes non-technical roles).

100%

Since participating in the hackathon, which of the following AI safety initiatives have you engaged in?

Please select all that apply. Please answer "Other" with any that are not covered here. You can provide context below.

Joined a research group

Contributed to an AI safety project (besides the hackathon project)

Continued working on my hackathon project

Attended a conference related to AI safety (NeurIPS, EAG...)

Joined a summer camp or workshop in AI safety (ML4G, Cooperative AI Summer School...)

Accepted or participated in a fellowship or training program (SERI MATS, CAIS Philosophy Fellowship...)

Read relevant literature

None

Other:

How much was the above attributed to participating in the hackathon?

What is the chance that you would have attended or done this if you had not participated in the hackathon. E.g. 70% if you had a low chance of reading up on literature unless you were motivated by a hackathon.

100%

Clear selection

Please describe any context for the two previous questions

E.g. which camps did you attend and how did the hackathon help you enter them, what did you read, and how did the hackathon help here? If you answered "None" above, you're very welcome to describe why or if the hackathon had a negative effect.

AI safety perspective shifts

Please describe any specific ways in which your understanding or approach to AI safety changed as a result of participating in the hackathon(s). We are very curious to hear concrete mindset-shifts about AI safety (e.g. "before I believed X, now I believe Y").
*

What skills did you improve during the hackathon?

We're curious to hear what you have experienced the most improvement in. Check all that apply.

Research teamwork

Networking and new friends in AI safety

Technical engineering in AI safety research

Development operations (such as setting up Github repositories etc.)

Understanding of the theory underlying AI safety

Machine learning engineering

Longtermism and global strategy

Other:

What did you learn?

To what extent do you believe your skills in areas relevant to AI safety (e.g., programming, problem-solving, collaboration) improved during the hackathon?
*

Not at all

To a large extent

Can you describe any moments or aspects of the hackathon(s) that were especially enjoyable or beneficial?

Here, we are interested in understanding which parts of hackathons provide you with the most value and we are curious to hear your specific experiences.

How your personal relation to AI safety was affected by the hackathon

Please answer the following questions with the most influential hackathon in mind.

Mindset shift

How has your perspective changed on AI safety?

Not at all

To a large degree

Clear selection

Do you feel more or less motivated for long-term engagement with AI safety?

5 is neutral.

Much less motivated

Much more motivated

Clear selection

How has your estimate of risk changed as a result of participating in the hackathon?

Negative indicates that you are much more pessimistic about our ability to solve problems in AI risk. Positive indicates that you are much more optimistic about our ability to solve problems in AI risk. 5 is neutral.

Very negatively

Very positively

Clear selection

Networking and Collaboration

What is the number of professional relationships formed during the hackathon, which could lead to future collaborations in AI safety?

How well do you feel you can contribute to solving problems in AI risk after the hackathon vs. before the hackathon?

5 is neutral.

Much less

Much more

Clear selection

Compared to what you would be doing otherwise, how much more did you get out of participating in the hackathon?

This is to assess if our hackathons provide you with counterfactual value.

<0.5x (less valuable than the alternative)

0.5x-2x (you got OK value out of participating)

2x-10x (you got a lot out of participating)

10x< (this was a much more valuable use of your time than the alternative)

Clear selection

Clear form

Never submit passwords through Google Forms.

This form was created inside of Apart Research. Report Abuse

Forms