Artificial Swarm Intellgience In The Context Of Singularity

Technical singularity is defined as a hypothetical future of superhuman machines with a cognitive capability far beyond the capacity of human minds. In the journey toward this potential technology revolution is something that I have been focused on called artificial swarm intelligence. A starling murmuration, something that people have told me is awe-inspiring, is a marvel of nature similar to an army of ants or a swarm of bees. How do all these individual entities organize around a common mission that includes a form of collaboration and unified orchestration as a team?

When thinking about swarms of AI bots or even nanobots, the foundational concept we want to define is what exactly AI bot are. They are software entities capable of machine learning, cognitive computing intelligence, behavioral analysis, understanding the ontology of things and capable of detecting entity state. Basically, they are highly intelligent software entities on the road to singularity.

In a similar paradigm to starling murmurations or swarms of bees, there is an emerging AI technology called swarm intelligence. When we think about a swarm in the context of bots, we often think of a botnet, physical bots (robots) or drone bots. Criminals are evolving botnets in the cybersecurity world faster than defensive botnets or protection methods are being developed.

Bots are becoming more common today, as we have chatbots, marketing bots, botnets and other forms of intelligence software entities that are associated with AI. However, they are thought of as singular entities. In the next emerging generation of bots, I believe they will act in a unified group or swarm. Think of a swarm of humans gathered together to collaborate and team up to build a house, including specialists such as plumbers, electricians, framers and roofers to unify and construct the home. In the emerging swarm intelligence, we will have specialized bots that can group together to accomplish similar orchestrated missions.

AI is being accelerated by the cost-efficiency of cloud computing in big data and with machine learning algorithms becoming a now critical mass. Traditionally, an evil botnet is a network of compromised hosts that are managed by one or more centralized servers. In the cybersecurity world, we call these command and control servers. Both Palo Alto Networks and Fortinet (FortiGaurd Labs) have defined swarms as Hivenets and Swarmbots. These botnets can even target industrial control systems such as distributed water systems for smart cities.

Previously in botnets, individual decisions were centralized and were not coordinated between the members of the swarm. There was no notion of voting or distributed sharing of communication resulting in consensus behavior. This evolution of machine learning in AI is evolving quickly and becoming more accessible to lower-level technical people. For example, in machine learning, there is the concept of ensemble learning, which can take the outputs of previous machine learning models and combine or fuse the results to achieve greater prediction capabilities.

Even more recent is how OpenAI has trained a group of AI agents to play hide and seek, using reinforcement learning to provide feedback loops into their learned intelligence. The AI agents even surprised the researchers who created them with new ways to win the game. Similar to the OpenAI AI agents, this AI goal for software to discover emergent patterns of knowledge and emergent unanticipated patterns of behavior.

Traditionally, swarm intelligence has been focused on spatial and temporal awareness in the physical world — think Battle Bots or the Terminator. This was displayed in the swarm of drones in the opening of the 2018 Winter Olympics in South Korea, which even concerned military officials. Like pixels in a 3-D image, each drone is assigned to act as an aerial pixel, filling in the broader canvas. There are other defining works in this spatial area, such as flocking algorithms, ant colony-optimization algorithms and collision-avoidance algorithms. Swarm robotics is a combination of AI and blockchain technology that enables an evolved form of collective communication between members of a botnet using more of a decentralized ledger approach.

In the end, what we should really strive for is a swarm of humans and bots (machines) that can work as a collective team. In the early days of AI this was referred to as a human in the loop — a collaborative squad that rallies around a common mission or orchestrated goal. For example, the company Unanimous AI has proven that swarms of humans are very effective at predicting future events and problems.

However, some of my industry peers suggest putting guardrails around AI when machines are involved. For example, in a human-bot swarm, there should be a moral compass. Are humans in control of the swarm? Are bots in control of the swarm, and humans have input but are just “followers”? If the bots and humans disagree on their opinions and decisions, who wins?

We already see this with Alexa and other voice-controlled assistants, where humans talk to software through chatbots using speech and not through traditional web consoles. The concept of singularity and swarm intelligence is one that is evolving rapidly in the back rooms of research centers and is something I am looking forward to seeing as new emerging technology innovation.