The next big wave in artificial intelligence innovation is at hand: agents.
With agents, the usefulness of AI will increase exponentially and enable businesses and governments to streamline their operations while making them more dependable, efficient and adaptable to circumstance, according to Satya Nitta, co-founder and CEO of Emergence, the futuristic New York-based computer company.
These are the first AI systems that can speak with humans and each other conversationally, which may reduce some of the anxiety people feel about AI — this unseen force set to transform our world. These agents use AI to perceive their environment, make decisions, take actions and achieve goals autonomously, Nitta said.
The term “situational awareness” could have been created for agents because that is the key to their effectiveness.
For example, an autonomous vehicle needs a lot of awareness to be safe and operate effectively. It needs every bit of real-time knowledge that a human driver needs on the roadway, including scanning traffic on all sides of the vehicle, looking out for an approaching emergency vehicle or a child who might dash into the road, or sensing a drunk driver.
Emergence is a well-funded startup aiming to help big companies and governments by designing and deploying agents for their most complex operations. It is, perhaps, easier to see how an agent might work for an individual and then extrapolate that for a large system, Nitta suggested.
Take a family vacation. If you were using an agent to manage your vacation, it would have to have been fed some of your preferences and be able to develop others. With these to the fore, the agent would book your trip, or as much of it as you wished to hand to the agent.
The agent would know your travel budget, hotel preferences and the amusements that would interest your family. It would do some deductive reasoning to allow for what you could afford and balance that with what is available. You could discuss your itinerary with the agent as though it were a travel consultant.
Nitta and Emergence are designing agents to manage the needs of organizations, such as electric utilities and their grids, and government departments, like education and healthcare. Emergence, along with several other AI companies and researchers, has signed a pledge not to work on AI for military applications, Nitta said.
Talking about agents built on open-source Large Language Models and Large Vision Models, Nitta said, “Agents are building blocks which can communicate with each other and with humans in natural language, can control tools and can perform actions in the digital or the physical world.”
Nitta explained further, “Agents have some functional capacity. To plan, reason and remember. They are the foundations upon which scalable, intelligent systems can be built. Such systems, composed of one or more agents, can profoundly reshape our ideas of what computers can do for humanity.”
This prospect inspired the creation of Emergence and caused private investors to plow $100 million in equity funding into the venture and lenders to pledge lines of credit of an additional $30 million.
Part of the appeal of Emergence’s agents is that they will be voice-directed and you can talk to them as you would to a fellow worker or employee, to reason with them, perhaps.
Nitta said that, historically, there have been barriers to the emergence of voice fully interfacing with computing. And, he said, there has been an inability of computers to perform more than one assignment at a time. Agents will overcome these blockages.
Nitta’s agents will do enormously complex things like scheduling the inputs into an electricity grid from multiple small generators or calculating weather, currents and the endurance of fishing boats and historical fish migration patterns to help fishermen.
At the same time, they will be adjusting to changes in their environment, say, for the grid, a windstorm, or the fish are turning south, not east, as expected, or if the wholesale price of fish has dropped to change the economics of the endeavor.
To laymen and to those who have been awed by the seeming impregnable world of AI, Emergence and its agent systems are reassuring because you will be able to talk to the agents, quite possibly in colloquial English or any other language.
I feel better about AI already — AI will speak English if Nitta and his polymaths are right. AI, we should talk.
Leave a Reply