The year is 2030. REST API is dead. Invoking requests results in your web browser inbuilt LLM guessing what is supposed to be returned from the server. When opening github.com you occasionally see cheese, at times animal shelter hotline.
How about submitting a prompt for a REST request and then the LLM decides what information you are allowed to access and what actions you are allowed to take? And it works like a batch job of sorts. One request but it can enter many flows.
> Failure-oblivious computing "enables servers to execute through memory errors without memory corruption. Our safe compiler for C inserts checks that dynamically detect invalid memory
accesses. Instead of terminating or throwing an exception, the generated code simply discards invalid writes and manufactures values to return for invalid reads, enabling the server to continue its normal execution path.
From the conclusion:
> Our results show that failure-oblivious computation enhances availability, resilience, and security by continuing to execute through memory errors while ensuring that such errors do not corrupt the address space or data structures of the computation. In many cases failure-oblivious
computing can automatically convert unanticipated and dangerous inputs or data into anticipated error cases that the program is designed to handle correctly. The result is that the program survives the unanticipated situation, returns back into its normal operating envelope, and continues to satisfy the needs of its users.
I wrote a similar program using Ruby metaprogrammming, but instead if a function is called that doesn't exist (say in tests) it has the LLM fix it dynamically
Wait till it gets gobbled by the next gen training data, and embedded in the weights of upcoming LLMs. Paired a clueless vibe coder, with no token limits.
The year is 2030. REST API is dead. Invoking requests results in your web browser inbuilt LLM guessing what is supposed to be returned from the server. When opening github.com you occasionally see cheese, at times animal shelter hotline.
How about submitting a prompt for a REST request and then the LLM decides what information you are allowed to access and what actions you are allowed to take? And it works like a batch job of sorts. One request but it can enter many flows.
Nightmare fuel
Relevant, from 20 years ago, "Enhancing Server Availability and Security Through Failure-Oblivious Computing", Rinard et al.: https://people.csail.mit.edu/rinard/techreport/MIT-CSAIL-TR-...
From the abstract:
> Failure-oblivious computing "enables servers to execute through memory errors without memory corruption. Our safe compiler for C inserts checks that dynamically detect invalid memory accesses. Instead of terminating or throwing an exception, the generated code simply discards invalid writes and manufactures values to return for invalid reads, enabling the server to continue its normal execution path.
From the conclusion:
> Our results show that failure-oblivious computation enhances availability, resilience, and security by continuing to execute through memory errors while ensuring that such errors do not corrupt the address space or data structures of the computation. In many cases failure-oblivious computing can automatically convert unanticipated and dangerous inputs or data into anticipated error cases that the program is designed to handle correctly. The result is that the program survives the unanticipated situation, returns back into its normal operating envelope, and continues to satisfy the needs of its users.
I wrote a similar program using Ruby metaprogrammming, but instead if a function is called that doesn't exist (say in tests) it has the LLM fix it dynamically
Don't leave us hanging!
Haha I will dig it up sometime, but it was a little prototype!
Beautiful. Unfortunately, I don’t think I can afford the Black Friday price unless I find a volunteer.
This is a delightfully horrible idea. Well played.
Or have LLM catch these errors during development
I wonder how many people will use it unironically despute the writing. It's probably going to be a nonzero number, right?
Wait till it gets gobbled by the next gen training data, and embedded in the weights of upcoming LLMs. Paired a clueless vibe coder, with no token limits.
only if your production environment is a raspberry pi under your bed haha
“Depending on what the AI thinks is funny today.”
lol