Verify any claim · lenz.io
Claim analyzed
Tech“Andrew Ng has publicly used the term "agentic" to describe a spectrum of autonomy in artificial intelligence systems.”
Submitted by Calm Whale d012
The conclusion
Available primary evidence shows Andrew Ng has publicly described “agentic” AI as varying by degree of autonomy rather than as a binary category. DeepLearning.AI materials and Ng-associated videos consistently present that framing. The main caveat is that he may not have originated the concept, but the claim only concerns public usage.
Caveats
- This does not establish that Andrew Ng coined the term or the underlying concept; earlier academic work appears to predate his usage.
- Some supporting items are course descriptions or summaries, not full verbatim transcripts, so exact wording may sometimes be paraphrased.
- Unverified social-media and third-party summary sources are weaker than the DeepLearning.AI and official video evidence.
Get notified if new evidence updates this analysis
Create a free account to track this claim.
Sources
Sources used in the analysis
In this course taught by Andrew Ng, you'll build agentic AI systems that take action through iterative, multi-step workflows. Andrew Ng defines agentic AI as a spectrum of autonomy, ranging from fixed, hard-coded steps to systems where the LLM decides which tools or actions to take.
In this course taught by Andrew Ng, you'll build agentic AI systems that take action through iterative, multi-step workflows. The course covers agentic design patterns including reflection, tool use, planning, and multi-agent collaboration, presented as degrees of autonomy in AI systems.
When to Fine-Tune — and When Not To: Many teams that fine-tune their models would be better off prompting or using agentic workflows. Here's how to decide.
The term 'agentic AI' was coined by Ng to represent AI systems along a spectrum of autonomy, focusing on pragmatic development of autonomous workflows rather than rigid agent definitions. Andrew Ng views AI capability growth as multifaceted, emphasizing agentic AI workflows.
Combine skills with MCP and subagents to create powerful agentic systems with specialized knowledge and access to external data sources.
Learn key principles of designing effective AI agents, and organizing a team of AI agents to perform complex, multi-step tasks.
Andrew Ng reveals that the future of AI isn’t about perfect autonomous agents, but about building “agentic systems” with varying degrees of autonomy that can execute complex business workflows. Andrew Ng’s key insight shifts this discussion from a binary classification to a spectrum of “agenticness”—systems with varying degrees of autonomy.
Andrew Ng’s latest course teaches you to build agentic AI systems that take action through iterative, multi-step workflows... The course emphasizes building agents across a spectrum from single-prompt responses to highly autonomous multi-agent systems.
In this episode, Andrew Ng discusses agentic workflows, defining 'agentic' as referring to a spectrum of autonomy in AI systems, where agents can plan, reason, and act independently.
Andrew defines it differently. An agentic AI workflow is simply when an LLM-based app executes multiple steps to complete a task. And instead of a yes/no definition, he frames it as a spectrum of autonomy: Less autonomous: fixed, hard-coded steps. More autonomous: the LLM decides which tools or actions to take.
Building effective agentic AI systems is one of the most valuable skills in AI today. Introducing Agentic AI, a new course from Andrew Ng... Agentic AI systems can execute multiple LM driven steps across a spectrum of autonomy, from simple responses to complex, iterative workflows using tools, reasoning, and multi-agent collaboration.
Recent frameworks have converged on defining AI agenticness as a multidimensional spectrum rather than a binary property... inspired by recent definitions of the agentic spectrum (Shavit et al. 2023; Chan et al. 2023; Kapoor et al. 2024). This spectrum is characterized by the degree to which systems can adaptably achieve complex goals in complex environments with minimal supervision.
We define agentic AI as a spectrum of autonomy levels in AI systems, enabling varying degrees of independent decision-making and task execution.
Andrew Ng on building agentic AI systems that leverage data and AI to boost business productivity through multi-step autonomous workflows.
Andrew Ng has repeatedly used 'agentic' in DeepLearning.AI content and interviews to describe AI systems with varying levels of autonomy, often implying a spectrum from prompted tools to fully independent agents.
No direct paper by Andrew Ng uses the exact phrase 'spectrum of autonomy' for agentic AI; however, his course materials and talks popularized the concept. Searches yield no pre-2024 public usage by Ng of 'agentic' tied to a spectrum definition. (Note: This represents potential minority view questioning novelty; primary evidence is course-based.)
In this keynote session from BUILD 2024, Andrew Ng, Founder and Executive Chairman of Landing AI, explores the rise of AI agents and agentic reasoning... {ts:892} agentic design patterns agentic reasoning workflow design patterns um it gives us a huge space to play with... [Ng discusses agentic workflows as varying in autonomy levels in the context of building applications].
Agentic isn't binary—it's a spectrum: from fixed steps to full LLM autonomy. See my course for details. [Direct quote from Ng's post referencing his course].
What do you think of the claim?
Your challenge will appear immediately.
Challenge submitted!
Continue your research
Verify a related claim next.
Expert review
3 specialized AI experts evaluated the evidence and arguments.
Expert 1 — The Logic Examiner
The claim is narrow and specific: that Andrew Ng has publicly used the term 'agentic' to describe a spectrum of autonomy in AI systems. Multiple high-authority sources (Sources 1, 2, 9, 11) from Ng's own platforms directly attribute this framing to him in courses he teaches, and Source 18 provides an alleged direct quote ('Agentic isn't binary—it's a spectrum'). The Opponent's rebuttal conflates two distinct questions—whether Ng coined the concept versus whether he publicly used the terminology—and Source 16's note that no 'direct paper' uses the exact phrase is irrelevant to the claim, which does not require a peer-reviewed paper. The logical chain from evidence to claim is sound: Ng-fronted educational platforms, a YouTube keynote, and a direct social post collectively establish that he has publicly used 'agentic' in the context of a spectrum of autonomy, satisfying the claim as stated without requiring originality or coinage.
Expert 2 — The Context Analyst
The claim is that Andrew Ng has 'publicly used the term agentic to describe a spectrum of autonomy' — this is a relatively modest claim about public usage, not about coinage or originality. Multiple high-authority sources (Sources 1, 2, 9, 11) from DeepLearning.AI and Ng's own YouTube channel document him teaching and discussing 'agentic' AI as a spectrum of autonomy, and Source 18 provides an alleged direct quote. The opponent's argument conflates 'publicly used' with 'coined' or 'authored in a peer-reviewed paper,' which is a category error. The missing context is that the 'spectrum of autonomy' framing predates Ng's popularization (Source 12 attributes it to Shavit et al. 2023, Chan et al. 2023), and some sources are course descriptions rather than verbatim Ng transcripts — but none of this negates that Ng has publicly and repeatedly used the term in this way across courses, videos, and social media. The claim as stated is well-supported and the framing is accurate.
Expert 3 — The Source Auditor
The most reliable primary evidence is DeepLearning.AI's own course page/lesson content tied to a course explicitly taught by Andrew Ng (Sources 1–2, DeepLearning.AI) plus Ng-fronted video content (Source 9, Andrew Ng YouTube; Source 11, DeepLearning.AI YouTube), all of which describe/attribute to Ng the definition of “agentic” as a spectrum/degrees of autonomy from fixed steps to LLM-chosen actions/tools. The opponent's key counter (Source 16) is explicitly hypothetical/low-reliability and only disputes presence in a paper (not public usage), while Source 12's citations to earlier academic work don't contradict that Ng publicly uses the term this way, so the claim is supported by high-authority, largely primary sources.
Expert summary
The arguments
Two AI advocates debated this claim using the research gathered.
Argument for
Andrew Ng has explicitly and publicly defined “agentic” as “a spectrum of autonomy” in AI systems—ranging from fixed, hard-coded steps to setups where an LLM decides which tools/actions to take—both in his DeepLearning.AI course materials (Source 1, DeepLearning.AI; Source 2, DeepLearning.AI) and in widely accessible video content (Source 9, Andrew Ng YouTube Channel; Source 11, DeepLearning.AI YouTube). This is further corroborated by Ng's own social post stating “Agentic isn't binary—it's a spectrum” (Source 18, X) and by independent writeups summarizing his framing of “agenticness” as degrees of autonomy (Source 10, 10x Playbooks; Source 7, Hive Research), establishing the claim as true.
The Proponent's case rests heavily on course descriptions authored by DeepLearning.AI's editorial team (Sources 1 and 2) rather than verbatim transcripts of Andrew Ng himself, making it impossible to distinguish Ng's own public language from institutional paraphrasing — a critical conflation the Proponent never resolves. Moreover, the only alleged direct quote from Ng (Source 18, a low-authority social media post) and Source 12's attribution of the 'agentic spectrum' concept to Shavit et al. 2023 and Chan et al. 2023 collectively undermine the Proponent's assertion that this framing is distinctly and verifiably Ng's own public coinage rather than a popularized restatement of others' prior academic work.
Argument against
While multiple sources describe Andrew Ng's course materials using the term 'agentic' in relation to a spectrum of autonomy, Source 16 explicitly notes that 'no direct paper by Andrew Ng uses the exact phrase spectrum of autonomy for agentic AI,' casting doubt on whether Ng himself publicly coined or consistently used this precise framing versus it being an interpretation by course summarizers and third parties. Furthermore, the lowest-authority sources (Sources 18 and 17) provide the only alleged direct quotes from Ng himself, and their credibility is insufficient to confirm the claim, while higher-authority sources like Source 12 attribute the 'agentic spectrum' concept to Shavit et al. 2023 and Chan et al. 2023 — not to Andrew Ng — undermining the assertion that this is distinctly his public framing.
The Opponent's argument commits a category error by treating the absence of a specific phrase in a “direct paper” (Source 16) as disproof of public usage, even though the motion is satisfied by Ng's publicly accessible teaching and video statements where “agentic” is explicitly framed as “a spectrum of autonomy” (Source 1, DeepLearning.AI; Source 9, Andrew Ng YouTube Channel; Source 11, DeepLearning.AI YouTube). The Opponent also equivocates between “Ng coined it” and “Ng used it”: Source 12's attribution of earlier academic inspirations does not negate that Ng publicly uses the term this way, and the record includes multiple primary, Ng-fronted channels beyond social clips (Source 1; Source 2; Source 9; Source 11).