Meta CEO Mark Zuckerberg recently said there could possibly be more AI agents in the world than humans.
While speaking to Rowan Cheung in a podcast, he said there are millions of small businesses in the world, and in the future all of them could have AI agents carrying out some functions for the company, like customer support and sales.
“I think every business – just like they have an email address, a website, and a social media presence today – will have an AI agent that their customers can talk to in the future,” he said.
AI agents will not just be limited to businesses, but content creators too could have their own AI agents. He said there are over 200 million people on Meta’s social media platforms who consider themselves as creators.
However, they struggle with limited time to engage with their communities, which desire more interaction. According to Zuckerberg, a potential breakthrough could involve integrating social media data with AI to reflect creators’ values and goals.
This could result in interactive, artistic artefacts and diverse AI agents tailored to individual needs and functions.
AI Agents for Everyone
In a recent interaction with AIM, Okta’s president for business operations, Eugenio Pace, said every employee in an organisation could have their own AI agents carrying out certain aspects of their day-to-day tasks.
On the consumer front too, there could be an AI agent built into your smartphone which could make a reservation, book a hotel or order a pizza on your behalf.
Venture capitalist Vinod Khosla predicts that most consumer interactions online will involve AI agents handling tasks and filtering out marketers and bots.
By 2025, the world is projected to have 7.4 billion smartphone users, which could potentially mean 7.4 billion AI agents. These AI agents need not be limited to smartphones; they would be accessible through your PCs or voice-activated virtual assistants like Alexa.
If this happens, Zuckerberg’s vision of more AI agents than humans would indeed come true. What’s more? Most individuals could have multiple AI agents that could be leveraged for different purposes, be it for business or personal use.
But do we have the technology to enable this?
Everyone is Building One
Earlier this year, OpenAI unveiled GPT-4o, demonstrating its impressive ability to converse almost indistinguishably from a human.
While OpenAI is yet to make the voice capabilities of GPT-4o available, according to reports, the startup is working on a new AI technology under the code name–Strawberry–to significantly enhance the reasoning capabilities of its AI models.
At Google I/O 2024, the tech giant unveiled Project Astra, a first-of-its-kind initiative to develop universal AI agents capable of perceiving, reasoning, and conversing in real-time.
In the same podcast, Zuckerberg reiterated that if every business were indeed going to have its own AI agent, he would want to be the enabler for these businesses.
Very recently, speaking to NVIDIA chief Jensen Huang at the ACM SIGGRAPH conference, Zuckerberg announced AI Studio, which helps create custom AI chatbots and characters.
Built with Llama 3.1, AI Studio, creators can build an extension of themselves or an AI agent based on certain interests.
That is not all. Hyperscalers like Microsoft, Google, and AWS have announced capabilities that can help enterprises build AI agents. Earlier this month, at the AWS New York Summit, the cloud provider announced that AI agents built through Amazon Bedrock would have enhanced memory and code interpretation capabilities.
Interestingly, there are many startups, which too are building AI agents for specific use cases. YC-backed Floworks recently announced ThorV2, which they claim is better than GPT-4o. Similarly, another Indian startup called KOGO is helping enterprises build AI agents in Indian languages.
Laying the Groundwork
Companies are not only developing AI agents but also laying the groundwork for their implementation. Meanwhile, others are discovering new business opportunities in the emerging field.
For instance, Pace revealed that Okta, which sells subscriptions to enterprises for its identity software, could provide the same for AI agents, thus significantly boosting their revenue.
Meanwhile, Beckn Protocol, which is an open specification that creates a common language for interoperability among different digital platforms and networks, could prove to be useful in the era of AI agents.
Sujith Nair, the CEO and co-founder of FIDE, earlier told AIM that a foundational contract structure will be needed when two AI agents communicate on humans’ behalf.
“This necessitates a programmable, machine-readable method of contracting in real-time, and achieving this requires an interoperable protocol like Beckn,” he said.
Beckn ensures there’s a verification process for these transactions when two AI agents (buyer and seller) are talking to each other.
