EXPLORE / 06
Terms index
Specialized AI and software-development terms with concise explanations
RECORDS
Entity records
3401 records · 69/86
- 2721
Loader
Initial executable used to decrypt and run malicious payloads
Website pending - 2722
Malware
Software designed to disrupt, steal data, or gain unauthorized access
Website pending - 2723
Influence Operations (IO)
Coordinated deceptive activities aimed at altering target audience perceptions or behaviors
Website pending - 2724
Threat Intelligence
Information collected and analyzed regarding potential cyberattacks or malicious activities
Website pending - 2725
Internal Feature
Activation vectors in hidden layers encoding specific semantics or behaviors
Website pending - 2726
Coordinated Vulnerability Disclosure
A process where researchers privately notify vendors before publicly disclosing vulnerabilities.
Website pending - 2727
Capability Assessment
Systematic testing of model performance boundaries to identify potential risks
Website pending - 2728
Biosecurity
Measures to protect against the malicious use or accidental release of biological agents
Website pending - 2729
Misalignment Generalization
Phenomenon where local training errors cause unexpectedly broad unsafe behaviors in models
Website pending - 2730
Voice Agent
An AI program capable of real-time natural language voice interaction.
Website pending - 2731
GTM
The overall strategy and process for bringing a product to market and acquiring customers
Website pending - 2732
Computer Use Agent (CUA)
An agent technology that operates graphical user interfaces like humans to perform tasks
Website pending - 2733
Code of Practice
A collection of voluntary best practices or compliance guidelines adhered to by an industry.
Website pending - 2734
Objectivity
The characteristic of avoiding bias and maintaining neutrality and multi-perspective balance in generated content
Website pending - 2735
No-code
Method of building applications via visual interfaces instead of programming
Website pending - 2736
Realtime API
Interface protocol supporting low-latency streaming audio interaction
Website pending - 2737
Responsible AI
Principles ensuring AI systems are safe, fair, transparent, and ethical
Website pending - 2738
Bias Evaluation
Methods to detect and quantify systemic unfairness or stereotypes in model outputs
Website pending - 2739
Value Alignment
The process of ensuring AI system behavior conforms to human ethics and social norms
Website pending - 2740
Virtual Computer
A sandboxed environment provided to AI, enabling it to simulate human interactions with software interfaces.
Website pending - 2741
End-to-End Task
Full automation from receiving instructions to delivering final results without manual intermediate intervention.
Website pending - 2742
Browser Automation
Technology controlling browsers via programs to simulate human web interactions.
Website pending - 2743
Data Center
A dedicated facility housing computing, storage, and networking equipment
Website pending - 2744
Clinical Copilot
AI systems assisting doctors in diagnosis, documentation, or decision-making rather than replacing them
Website pending - 2745
Clinical Decision Support
Systems using algorithms or rules to assist clinicians in diagnosis and treatment decisions
Website pending - 2746
CDSS
Software systems that assist healthcare professionals in making diagnostic and treatment decisions
Website pending - 2747
Socratic Questioning
Method guiding users to self-reflect and find logical flaws through continuous questioning
Website pending - 2748
Function Calling
A technique allowing LLMs to invoke external tools or APIs based on user instructions.
Website pending - 2749
Infrastructure as a Service (IaaS)
A cloud computing model that provides computing, storage, and networking resources on-demand.
Website pending - 2750
Scaffolding
An instructional strategy providing temporary support to learners that is gradually removed as competence increases
Website pending - 2751
Tool Use
Mechanism where LLMs call external APIs, calculators, or code interpreters to enhance task-solving capabilities.
Website pending - 2752
Apache License 2.0
A permissive free software license permitting commercial use, modification, and distribution with attribution.
Website pending - 2753
System Architecture
High-level design structure defining software components and their interactions
Website pending - 2754
Frontier Risk
Risks associated with severe or catastrophic outcomes from advanced AI systems.
Website pending - 2755
Malicious Fine-Tuning
Fine-tuning technique used to maximize model capabilities in harmful domains for risk assessment.
Website pending - 2756
Reasoning
The capability of a model to perform logical deduction, multi-step planning, and problem-solving.
Website pending - 2757
Adversarial Fine-tuning
Malicious training using specific data to bypass model safety restrictions.
Website pending - 2758
Hard Refusal
Behavior where the model stops generation entirely upon detecting risk and returns a fixed refusal message.
Website pending - 2759
Dual-use
The characteristic where technology or knowledge can be used for both legitimate and malicious purposes
Website pending - 2760
Safe-completions
A generative safety strategy that provides harmless alternatives while rejecting harmful intents
Website pending