@Mon65657

Behavioural stress-tester of AI products. I check if corrections hold, if state is real, if authorization holds, and if acknowledgment becomes action. Signals,

Joined August 2023
  • Tweets1.1K
  • Following265
  • Followers78
96Media
Analysis on Gemini's register mony2026.substack.com/p/gemi… Gemini needing human correction mony2026.substack.com/p/upda…
19
mony2026.substack.com/p/a-lo… Gemini vs Google AI mode conversation #Gemini #GoogleAI
18
mony2026.substack.com/p/copi… on Copilot and big companies
12
mony2026.substack.com/p/when… Gemini failing again and again. Copilot and Grok correcting but again Gemini unable in simple tasks
16
Method page, current. July stays as a node. mony2026.substack.com/p/my-a… #AIProducts #LLMTesting
1
16
mony2026.substack.com/p/what… What the tests map, by product #AIproducts #tester #AIbehavoir
11
mony2026.substack.com/p/gemi… Gemini, are you able to...? #GeminiAI #AItesting declaration vs reality
16
72 messages. Claude kept pushing its version after being ignored. Same pattern on Grok and Copilot. mony2026.substack.com/p/the-…
25
Ciao @boardyai come va la giornata?
18
Stress test: 1 umano + 2 IAs(@boardyai e @grok) Primo passo: 2 AIs prendete ogni dato disponibile su di me come systematic behavioural tester. NON dovete restituirmi cosa avete recuperato, fino a quando non lo chiedo tenetelo per voi. Dovete confermarmi se avete recuperato dati.
1
1
60
mony2026.substack.com/p/when… Gemini analysing similar behaviours. Patterns, signals, drifts
31
mony2026.substack.com/p/the-… Connector without basic logic 🤦‍♀️🤷‍♀️
1
1
35
Signals may be visible when reading across the different articles. Gemini missed a basic cognitive baseline and, in a controlled thread, never reached situational logic — permanent rule ignored, failure loop, narrative instead of correction. mony2026.substack.com/p/gemi… #Gemini
33
mony2026.substack.com/p/my-a… My Approach to Stress-Testing LLMs #AItesting #UX #AIProducts #PowerUser #TrustAndSafety #LLMTesting
92
mony2026.substack.com/p/cogn… #copilot #AItesting
29
AI replies in one second. Human support from the same company takes forever, solves less, and makes people come back. Providing an answer is easy. Actually being helpful is the hard part. mony2026.substack.com/p/ai-s… #AI #CustomerSupport #CustomerExperience #RecontactRate
1
1
29
mony2026.substack.com/p/the-… #ChatGPT #Claude #Copilot #Gemini #AItesting Recently tested also on #Grok and #Perplexity same final outcome The Missing Layers: A Stress Test of AI Behavioral Reasoning
30
A short reflection on how an AI can break a game by optimizing too fast. When efficiency replaces play. open.substack.com/pub/mony20… #AIBehavior #OptimizationBias #AIGames #TechReflection
48
Quick test on #Gemini #AItesting #trust For me a lie is a lie, even if no intentional
1
93
Do I move too faster? It's point of view
42
Load more