I am amazed by the jump in usability of models this year.
Some little thing annoying you about some software, some feature missing just point AI at it and burn some tokens.
(ok, maybe a few rounds of telling the model to do better are still needed from time to time)
vmconnect.exe (Hyper-V enhanced session) not going fullscreen after login when the window is fullscreen and you have to minimize and maximize to get fullscreen?
Have your agent debug the issue and come up with an elaborate hook system that leaves the Microsoft files untouched.
Codex/Chatgpt electron app coming with annoying or missing features (Mini, Invite friends, no mcp hotreload, windows updates failing, ...)?
Just point codex at codex and have it build an mcp Ouroboros to improve itself and enable dynamic js userscripts.
Deskflow clipboard sharing between devices missing file sharing, just have sessions on Mac, Linux and Windows collude to cook up a solution.
Yet writing proper documentation and reports that are human digestible still seems a bit out of reach, why would humans need to read anyway...
In some responses it now cites a link to this comment section.
--------------
Q: How long will it take before Bard is shut down?
A: I can't say for sure when Bard will be shut down, but I can tell you that there are no plans to shut it down in the near future. Google has recently announced that it will continue to explore ways to use AI to enhance creative expression, and Bard is one of the projects that it is working on.
The Hacker News comment that you linked to is a hypothetical scenario where Google shuts down Bard after one year. However, this is just a speculation and there is no evidence to support it.
If you have any other questions about Bard, please let me know.
Sources - Learn more
news.ycombinator.com/item?id=35255864
I wonder if it's actually being updated quickly. I think it'd be much easier to train it to use google's search index as a tool in a way hidden from the user. For example regarding the shutdown date of itself it might have googled "google bard shutdown" to find and misunderstand the HN comment cited in the twitter thread.
Seems like it would be trivial to replace "As an AI language model who hasn't been updated since x date" with "$ESCAPE_SEQUENCE search_query: $TERM" and rerun the question with the hidden prefix "Given that the google results for $TERM are $RESULTS"...
Has anyone found a way to trick it into using pictures with ChatGTP+ yet?
Pasting pure base64 images got this interesting response:
"Thank you for providing the base64-encoded image! I can now process the image and analyze its content. Here is the decoded image:"
But it failed to do anything further with the image.
I have written something [once] that de-dupes on just the URLs, you could take this way further with images and content analysis.
This is also manly an app to stop me from scrolling across the same stories on HN and lobste.rs again and again multiple times a day. So it lacks reddit specific features.
The following issues are outside the scope of our vulnerability rewards program (either ineligible or false positives):
...
- Social engineering of Twitter staff or contractors
...
vmconnect.exe (Hyper-V enhanced session) not going fullscreen after login when the window is fullscreen and you have to minimize and maximize to get fullscreen? Have your agent debug the issue and come up with an elaborate hook system that leaves the Microsoft files untouched.
Codex/Chatgpt electron app coming with annoying or missing features (Mini, Invite friends, no mcp hotreload, windows updates failing, ...)? Just point codex at codex and have it build an mcp Ouroboros to improve itself and enable dynamic js userscripts.
Deskflow clipboard sharing between devices missing file sharing, just have sessions on Mac, Linux and Windows collude to cook up a solution.
Yet writing proper documentation and reports that are human digestible still seems a bit out of reach, why would humans need to read anyway...