Note: The linked site is a personal project, and all views expressed here are my own.
TL;DR
I created an interactive flowchart about various scenarios how the future of AI might play out.
By setting various conditional probabilities you can see charts of what the resulting estimated probabilities for good, ambiguous, and existentially bad outcomes are.
Use the site as a conversation starter and tool for reflection. You can share your results either as images of the resulting outcome charts, or just by sharing your personalized URL (which contains your estimates).
Feel free to clone, expand, and modify the site. I’d love to see how other people would structure their flowchart!
Additional context (copied from the website itself)
Background
Many experts and CEOs of the leading AI companies think advanced AI could pose catastrophic, or even existential risks to humanity.
I think we should take these concerns extremely seriously, and work hard to ensure that AI is developed in a way that is safe and beneficial for humanity.
This map is an attempt to visualize some of the key questions and uncertainties that could determine the future of AI, and to explore how different assumptions about these questions could lead to different outcomes.
How to use
White nodes are questions; grey nodes are intermediate states; red/yellow/green nodes are bad/ambiguous/good outcomes.
Adjust the sliders to set the conditional probabilities for each question.
The map and charts will update in real-time to show the probabilities for each node and outcome category.
Adjust the settings to make more likely paths bolder, or less likely paths transparent. By setting the minimum opacity to zero, impossible branches of the map will become fully transparent.
Click on a node to set it as the root node for the probability calculations, i.e. “what happens if we assume we have reached this state?”. Click the same node again, or the “START HERE” node to reset the probability mass to the start.
For clicked and hovered-over nodes in the map, the corresponding sliders in the sidebar are highlighted in the same color, and vice-versa.
The number in each node represents the total probability (in %) to reach this node.
The numbers on edges represents the the chosen conditional probabilities for each question.
Caveats
The future is notoriously difficult to predict. I’m not an AI expert, and I’m sure this map is flawed in various ways. Any outcome probabilities computed here are highly speculative, and should be taken with a grain of salt.
Consider this map as a reflection and conversation tool for exploring your own assumptions, and comparing them with those from a person with a different intuition about the future of AI.
The sliders represent conditional probabilities, i.e. the probability for the given question assuming we reached the corresponding node in the graph.
Feedback
If you have any feedback about this map (both about the questions/structure or the user interface), please let me know via this Feedback Form.
Addendum
Originally, I had planned to work more on the site before sharing it here, including features like
Compare your estimates with that of another person (using their custom URL).
Submit your estimates towards a global aggregate forecast.
Add links for further reading to each node.
Make it mobile-friendly.
Add more detailed nodes about e.g. risks from multi-agent scenarios.
And much more...
However, I did notice that I do not really have the time & energy to work on this project further, which is why I’m sharing it now in its current state.
I do hope you find value in it. Also, in case you’d like to create a similar website, but with a different flowchart structure (or additional features), please do feel free to clone my code on Github. In order to update the flowchart structure, you basically just need to update the strings in the “graph.js” file; no coding needed!
It’s nicely done, and I think it will be helpful for anyone in refining their thinking. Thanks for doing this! And for making it easy to edit. That makes it a more general tool for scenario evaluation.
Gave it shot on my phone (Android, Firefox). Works great! Fun and easy to experiment with. I feel like there are a few key paths missing, or perhaps the framing of the paths doesn’t quite line up with my understanding of the world.
With the options I was given, I get 8.9% extinction, 1.1% good, remainder ambivalent.
Personally, I think we have something like a 10% chance of extinction, 88% chance of utopia, and 2% chance of ambivalent.
At some point I think carefully about the differences between my model and yours and see if I can figure out what’s going differently.
https://swantescholz.github.io/aifutures/v4/v4.html?p=7i85i0i50i30i0i100i38i26i14i12i1i31i68i61i21i7i9i17i78i0