Home > Robotics > eXtended Artificial Intelligence

eXtended Artificial Intelligence

We have all dreamt of an utopic world, with flying cars gliding in the air or with robots doing our bidding or one voice command activating gadgets and machines. What is the key to achieving such a world? Two words – Artificial Intelligence.

But we have also seen movies like ‘I, Robot’ where the machines turn against the very humans who created them. Wasn’t this a foreseen risk by giving super-machines unlimited knowledge and control? In a sense, machines evolve and learn very similarly to humans. We know this fact and capitalize on it through Machine Learning.

Artificial intelligence – artistic concept. Image credit: geralt via Pixabay (free licence)

What if computers evolve one day to function independently from humans?

The key to creating a successful and efficient artificial intelligence system is to fine-tune its interaction with humans. The ultimate goal of AI is to automate human life by outsourcing it to a machine. We are giving the computer control over small, day-to-day tasks that we expect to be carried out with the same decisiveness and method.

How to achieve this?

Carolin Wienrich and Marc Erich Latoschik from the Human-Technology-Systems Group at University of Würzburg  produced a research paper highlighting the nuances of human-AI interaction and how to create a human-centered AI model by considering various factors and ‘embodiments’ affecting human perception and, in continuation, how to investigate systematically such interaction between human and artificial intelligence.

The paper presents the subject on two fronts which are as follows:

1. It uses the XR-AI continuum framework to generate theoretical treatment and model of human-AI interaction.

2. This research paper presents two experiments as an example investigating a human-centric approach to two different AI systems.

The claim presented by scientists about the possibility of usingcomputers to replicate and amplify human intelligence led to the coining of the term ‘Artificial Intelligence’ by John McCarthy in 1956. Ever since a whole new forum of computer sciences and information technologies has been blown wide open. Scientists have been researching methods to conducively integrate AI into human life without any significant repercussions.

One of the most important aspects of this research was to analyze possible scenarios of social interaction between AI and humans. In the future, users of artificial intelligence will have different needs and use for it and would approach it differently. The basis behind this research was to comprehensively understand such nuances and come up with an AI model susceptible to human biases and perception.

This experiment was conducted using an XR design space. XR continuum framework is a robust design and a relatively new perspective used to represent AI interaction partners using several shapes and models, from simple non-animated machines to intelligent industrial mechanized robots and even human look-alikes without facing any physical engineering or manufacturing challenge.

The research team designed a testbed system for human-AI interaction, which has the following main features that make it an attractive platform to conduct human-AI research on, and the features are as follows:

1. Accessibility: This experimental space can directly capture the insight of various users as it has a high degree of adaptability.

2. Versatility: This system provides several experimental modules and simulations with various difficulty levels to gain an objective conclusion.

3. Tangible Training: The use of XR environment provides the opportunity to simulate even mis-usage scenarios, while also implementing the concept of explainable AI for assessment of consequences.

The conclusion of this experiment was drawn in correspondence with the simulated robot’s ability to socialize with humans of various age groups and genders. The following conclusions were drawn:

1. The non-complex and straightforward build of the robot was considered more friendly, human-like, and less risky to automate and use.

2. Conversational robotic systems were more preferred than the non-conversational ones.

3. Women found conversational robots more approachable than the non-conversational ones. Men showed no such differences.

4. Women also found the conversation with the AI robots heady and of a much higher quality than their male counterparts.

5. Men in general preferred AI robots more, whilewomen preferred only conversational robots.

While this model mainly highlights the physical interaction between man and machine, it can also be extended to cognitive biases and perceptive thinking at a later stage.

This research could potentially open doors to creating advanced human-AI interfaces. But the one major limitation of this research is that it was conducted on a virtual simulation platform and not in the real world. The interactions only resemble reality. Hence we cannot be completely sure about the reliability of results. Not only that, but also the current system is relatively very simple and would not take into consideration many factors that may arise in real-world interactions. Such limitations could promptly develop into huge drawbacks when the model is introduced to reality.

Another limitation of the model is the inaccessibility of XR testbeds for students, who are the most practical stakeholders for this research model.

The future of the artificial intelligence revolution heavily relies on the cohabitation of humans and AI-induced machines and how they work with and around each other. Research like this sets a basis for what could be the most efficient and comprehensive AI model ever created

Source: Carolin Wienrich, Marc Erich Latoschik, “eXtended Artificial Intelligence: New Prospects of Human-AI Interaction Research“.

Source

Subscribe
Notify of
0 Comments
Inline Feedbacks
View all comments
0
Would love your thoughts, please comment.x
()
x