Learn to say popular AI red-teaming and adversarial testing tool names correctly.
0 / 5 completed
1 / 5
How is Garak (open-source vulnerability scanner for probing weaknesses in large language models) correctly pronounced?
Garak is pronounced 'GAR-ahk' — stress on the first syllable, rhymes loosely with 'barack'. In a technical interview: "Garak ran a hundred known jailbreak probes against the model in a single automated pass."
2 / 5
How is PyRIT (Microsoft's Python Risk Identification Toolkit for testing generative AI systems) correctly pronounced?
PyRIT is pronounced 'PY-rit' — 'Py' (like Python) plus 'rit', rhymes with 'writ'. In a technical interview: "PyRIT automated the red-teaming conversation, escalating each attack strategy until one succeeded."
3 / 5
How is Counterfit (Microsoft's open-source tool for security testing of AI and machine learning systems) correctly pronounced?
Counterfit is pronounced 'KOWN-ter-fit' — echoing the everyday word 'counterfeit', stress on KOWN. In a technical interview: "Counterfit generated adversarial examples that fooled our image classifier almost every time."
4 / 5
How is TextAttack (Python framework for generating adversarial attacks against NLP models) correctly pronounced?
TextAttack is pronounced 'TEKST-uh-tak' — 'text' plus 'attack', both plain English words. In a technical interview: "TextAttack swapped a handful of words and flipped the sentiment model's prediction completely."
5 / 5
How is ART (Adversarial Robustness Toolbox, IBM's open-source library for defending machine learning models) correctly pronounced?
ART is pronounced 'AY-AR-TEE' — every letter spoken individually, A-R-T, short for Adversarial Robustness Toolbox. In a technical interview: "ART measured how much the model's accuracy dropped under each adversarial perturbation."