Human-like Artificial Intelligence (Human-like AI) has a long and storied history.
At its core, the pursuit of human-like Artificial Intelligence boils down to humanity’s innate need to understand what it means to be human. Our creations of human-like AI, including the likes of chatbots, virtual beings and virtual human agents, stretch the boundaries of what humankind can create, along with raising the question of how far we should go.
In our Connectome World portal, five different categories of human-like AI will be explored:
In this post, we will explore the definitions to the five categories of human-like artificial intelligence.
Human-like AI #1: Chatbot
According to the Oxford Dictionary, a chatbot is “a computer programme designed to simulate
Interactions with chatbots are usually limited, as they are unable to deal with the quirks of natural language and require very specific and well-defined decision tree systems. They are meant to be used independently from human operators.
The first chatbot, ELIZA, was created in 1964 by Joseph Weizenbaum. ELIZA was an example of an early natural language processing programme and while it could engage in conversations with human beings, it
The next great leap forward for chatbots was A.L.I.C.E., created in 1995. Its language processing abilities won awards for its sophistication. The year 2001 saw the introduction of SmarterChild, a chatbot developed for use on instant messaging networks.
In many ways, ELIZA, A.L.I.C.E., and SmarterChild laid the foundations for modern chatbots that all build upon the ground-breaking work of their predecessors.
Human-like AI #2: Virtual Being
A virtual being is defined as “a character you know isn’t real but with whom you can build a two-way emotional relationship”. Basically, it refers to any online persona that isn’t real but displays enough human characteristics that it can form a personal relationship with a human being.
While the concept of a being that exists only in the virtual instantly brings to mind virtual reality, virtual beings can exist outside of virtual reality. At their core, virtual beings are meant to replicate the life of a human, except that they’re living in the virtual realm.
Some can communicate like a chatbot, while others exist solely as characters in video games. Some have jobs, like Lil Miquela, and others, like
Even though the definition of what a virtual being is can seem so broad it loses any real meaning, the key to what defines a virtual being is that it must be realistic enough to form a relationship with humans.
If one wants to consider a timeline of virtual beings, then the first virtual being can be described as the first character in an offline or online video game, movie, television series or programme that an individual first formed a personal relationship with, a relationship on the same level as a friend or lover.
Human-like AI #3: Virtual Assistant
Like chatbots, virtual assistants (VAs) leverage natural language processing to create their interactions. What separates virtual assistants from chatbots, however, is their level of sophistication. VAs are more dynamic and natural than chatbots, and they can often perform a wider variety of tasks without sacrificing accuracy.
Virtual assistants can often be used across multiple platforms while being able to maintain consistency and accuracy in their tasks. Some virtual assistants handle simpler tasks like scheduling meetings or sending emails while others can create automatic notifications
The only things that make them different from virtual human agents (VHA) are that they lack the body and the sophistication of a VHA. A key capability for all virtual assistants is the ability to recognise voice commands.
The origins of virtual assistants began in the 1960s and 1970s with the creation of the IBM Shoebox and the Carnegie Mellon Harpy Programme. They were the first programmes that supported voice capabilities. Fast forward to the 1990s, and we saw the introduction of Microsoft’s Clippy and Dragon Dictate. The 90s was also when the term, “virtual assistants” was coined. The rest, as they say, is history with the 2000s seeing the introduction of first Siri, followed by Google Now, Alexa and Cortana.
Human-Like AI #4: Virtual Human Agent
Like chatbots and virtual assistants, Virtual Human Agents (VHAs) have natural language processing capabilities that allow for conversations. Unlike the previous two, VHAs also possess deep AI-powered systems that give them capabilities such as sight, hearing, emotions, and intelligence. They can react to images and situations in the real world, converse, operate, think, and communicate as a human would.
Virtual Human Agents tend to be “embodied entities” where users can see the avatar and communicate with them just like a real human. Unlike virtual assistants, the uses of virtual human avatars are much more varied. Some have been used as healthcare support staff while others can work as receptionists in a company.
The expanded capabilities of virtual human agents, however, have a downside. The sheer amount of data necessary and the increased functionality require more robust safeguards in place such as using blockchain technology to keep track of decisions made.
As a more advanced virtual assistant, Virtual Human Agents are a relatively new concept. One of the first Virtual Human Agents is named Rachel and was developed by Connectome based on Couger technology in 2018. While Couger’s technology has been around since 2006, it was only through recent advances in technology that true Virtual Human Agents could be developed.
Human-Like AI #5: Humanoid
In robotics, humanoids are generally defined as “a robot with a body shape built to resemble the human body”. There are many ways to design such a robot for functional purposes like interacting with human tools or to play games. Androids, robots built to aesthetically look like a human are considered a subclass of humanoids. With advancements in the field and increasing subdivisions for proper classification, general use of the term humanoids is growing increasingly rare with a move towards more specific terms such as androids
Unlike virtual beings, chatbots, virtual human agents and virtual assistants, the history of humanoids is almost as long as the history of human-like Artificial Intelligence. Discounting the myths and legends, one of the first humanoids would be Al-Jazari’s orchestra of mechanical men from 1206.
The first humanoid invention that can be verified more accurately would be Leonardo da Vinci’s Automa Cavaliere – Automaton Knight – which was invented in 1495. Honda’s ASIMO, which was created in 1986, would be the first significant robot in modern times.
Human-Like Artificial Intelligence: Where Do We Go From Here?
Are we ready for the next generation interface of human-like artificial intelligence? Will AI be seen as a trusted friend one day, instead of a mere tool?
Time will tell. Meanwhile, let’s stay tuned to Connectome World for more updates on human-like artificial intelligence! Read more interviews with AI thought leaders here.