Silicon Valley is bullish on AI agents. Sam Altman, Openai CEO, said agents This year will “join the workforce” this year. Microsoft CEO Satya Nadella predicted that agents will replace a certain knowledge job. Marc Benioff, Director General of Salesforce, said Salesforce’s goal is to be “the number one supplier of digital work in the world“By the various” agent “services of the company.
But no one can seem to agree with what AI -agent isexactly.
In recent years, the TE industry industry has boldly proclaimed that AI “agents” – the newest mouthpiece – will change everything. Just like AI talks like AAIP of OPIC Given us new ways surface information, agents will essentially change as we approach work, claim CEOs like Altman and Nadella.
That may be true. But it also depends on how “agents” are defined, which is not an easy task. Similar to other AI-related slangs (eg “multimodal”, “act” and “ai” themselves), the terms “agent” and “active” become diluted until nonsense.
This threatens to leave Openai, Microsoft, Salesforce, Amazon, Google and the countless other companies building entire product ranks around agents in an awkward location. Amazon agent is not the same as Google’s agent or any other seller, and this leads to confusion – and frustration of customers.
Ryan Salva, senior director of product at Google and ex-Github Copilot manager, said he came “hate” the word “agents.”
“I think our industry predominates the term” agent “to the point where it is almost meaningless,” Salva told Techcrunch in an interview. “[It is] One of my pets Peeves. ”
The dilemma of an agent’s definition is not new. In a piece last year, former Techcrunch reporter Ron Miller Asked: What is AI -agent? The problem he identified is that almost every company building agents approach the Te Techniko differently.
It’s a problem worsened recently.
This week, Openai released a blog post This has defined agents as “automatic systems that can independently perform tasks on behalf of users.” However in the same week, the company was released Developer documentation This defined agents like “LLMs equipped with instructions and tools.”
Leher Pathak, Openai’s API Production Marketing, later said in A Post on x that she understood the terms “helpers” and “agents” to be interchangeable – further putting the waters.
Meanwhile, Microsoft blogs Try to distinguish between agents and AI helpers. The first, which Microsoft calls the “new programs” for “AI-operated world”, can be tailored to have particular expertise, while helpers simply help with general tasks, such as editing emails.
AI -Lab -Anantropic deals with the hodgepodge of agent definitions a little more directly. In a a Blog postAntropic says that agents “can be defined in several ways”, including “fully autonomous systems that work independently during extended periods” and “prescribed implementations that follow predefined workflows.”
Salesforce has what is perhaps the most comprehensive definition of AI “agent.” According to the giant software, Agents are “A kind of […] A system that can understand and respond to customer surveys without human intervention. “The company’s website lists six different categories ranging from” simple reflex agents “to” useful -based agents. ”
So why the chaos?
Well, agents – like AI – it’s a cloudy thing, and they are constantly evolving. Openai, Google and Perplexity have just started sending what they consider as their first agents – Opennai operator,, Google’s Mariner projectand A shopping agent of perplexity – And their abilities are through the map.
Rich Villars, GVP of Global Research at IDC, noted that ie companiesnological companies “have a long history” by not rigidly joining technical definitions.
“They care more about what they are trying to accomplish” on a technical level, Villars told Techcrunch, “especially in fast-developing markets.”
But marketing is also largely to blame, according to Andrew NG, the founder of AI -Learning Platform Deeplearning.ai.
“The concepts of AI ‘agents’ and ‘agent’ workflows used to have technical significance,” NG said in a recent interview, “but about a year ago, marketers and some large companies took possession of them.”
The lack of a unified definition for agents is both opportunity and challenge, Jim Rowan, head of AI for Deloitte, says. On the one hand, ambiguity allows flexibility, leaving companies to set agents according to their needs. On the other hand, it may – and probably already have – lead to “misaligned expectations” and difficulties in measuring the value and ROI of agent projects.
“Without a standardized definition, at least in an organization, it becomes difficult for reference activity and ensuring consistent results,” Rowan said. “This can result in various interpretations of what AI agents must deliver, perhaps complicating project goals and results. Ultimately, while flexibility can cause creative solutions, a more standardized understanding would help businesses better navigate the AI -agent landscape and maximize their investments.”
Unfortunately, if the revealing of the term “ai” is some indication, it seems unlikely that the industry will join around one definition of “agent” ever soon – if ever.
