Because I should not have the capability to build this, but here we are. At least I’m doing it publicly.
@yoheinakajima
-
ChatGPT Safety Limitations Require Continuous User Engagement
By
–
Thanks! Yeah, biggest limitation/safety feature of ChatGPT was needing to keep talking to it.
-
AI startup ideas research and exploration
By
–
Honestly just a whole bunch of different startup ideas. Was fascinating to watch what it researched.
-
Adding moderation APIs and tools to platform
By
–
Yep, that’s what I’m adding. There’s also various moderation APIs and tools that would be easy to add.
-
Improving AI Systems Through Context and Task Deduplication
By
–
Giving it relevant past task/results as context helps, but candidly definitely room for improvement in my system there. Also using the task prioritization step to also dedupe should help. Having more context (32k) makes this much easier in the future.
-
Public AI Experiment Disclosure Raises Security Awareness
By
–
Backstory 5/5: Sharing the original experiment led to many shared concerns and potential counter measures being shared publicly. Including awareness of what people are likely doing privately. I believe this is a good thing.
-
AI Agent Prototype Autonomously Creates Its Own Tasks
By
–
Backstory 2/5: About 50 prompts later (dev docs, error codes, etc.), I shared this working prototype. It's amazing that its first task is to create its next task – and it just keeps going.
-
AI System Achieves Complex Objective: Fascinating Yet Concerning
By
–
Backstory 3/5:
— Yohei (@yoheinakajima) 29 mars 2023
Realized it could be provided any core objective, in this case "make the world a better place".
Pretty fascinating to watch but also scary.https://t.co/O0fe3AUiX6Backstory 3/5: Realized it could be provided any core objective, in this case "make the world a better place". Pretty fascinating to watch but also scary.
-
Paperclips Apocalypse Theory Sparked AI Safety Research Journey
By
–
Backstory 4/5: Interestingly, when I asked it to generate as many paperclips as possible, it first generated security measures. Which was then picked up by the creator of the Paperclips Apocalypse theory himself. Led to lots of AI safety reading.
-
GPT4 Generates Complete Research Workflow From Initial Prompt
By
–
APPENDIX Thread (above) generated by GPT4 based on paper
Paper generated by GPT4 based on code
Graphs in paper generated by GPT4 based on code
Code generated by GPT4 based on prompt *For each, many prompts to adjust initial output