What if you don't need MCP at all?

(mariozechner.at)

Show context

jrm4 ◴[16 Nov 25 21:34 UTC] No.45948621[source]▶

Yeah, "MCP" felt like BS from jump. Basically it's the problem that will always be a problem, namely "AI stuff is non-deterministic."

If there was some certainty MCP could add to this equation that would perhaps be theoretically nice, but otherwise it's just .. parsing, a perhaps not "solved" problem, but one for which there's already ample solutions.

replies(1): >>45948883 #

1. chpatrick ◴[16 Nov 25 22:09 UTC] No.45948883[source]▶

>>45948621 #

Why are they nondeterministic? You can use a fixed seed or temperature=0.

replies(2): >>45949709 #>>45949732 #

2. datadrivenangel ◴[17 Nov 25 00:07 UTC] No.45949709[source]▶

>>45948883 (TP) #

The whole point of "agentic AI" is that you don't have to rigorously test every potential interaction, which means that even a temperature zero model may behave unexpectedly, which is bad for security.

3. urbandw311er ◴[17 Nov 25 00:10 UTC] No.45949732[source]▶

>>45948883 (TP) #

Even with zero temperature (and some of the latest models don’t allow that) you’re still not absolutely guaranteed deterministic output.

replies(3): >>45950196 #>>45951036 #>>45957887 #

4. ◴[17 Nov 25 01:50 UTC] No.45950196[source]▶

>>45949732 #

5. ares623 ◴[17 Nov 25 05:22 UTC] No.45951036[source]▶

>>45949732 #

Just repeatable output.

6. chpatrick ◴[17 Nov 25 20:26 UTC] No.45957887[source]▶

>>45949732 #

What does deterministic mean in this case?

replies(1): >>45980900 #

7. jrm4 ◴[19 Nov 25 15:37 UTC] No.45980900{3}[source]▶

>>45957887 #

Right, I'm using a perhaps loose definition of deterministic here, as in -- "With AI tools, you cannot 100% predict any output based on input, the way you can with most programming."

replies(1): >>45982862 #

8. chpatrick ◴[19 Nov 25 18:16 UTC] No.45982862{4}[source]▶

>>45980900 #

It's also kind of an impossible request given that AI tools are for generic unpredictable input.

↑