Categories: robots

Han and Sophia can recognize human expressions. Can they have a real conversation, too?

Evolved on one hand, but unconvincing on the other

Han

Hanson Robotics, a company founded by David Hanson, has developed a humanoid robot named Han, able to recognize facial expressions, maintain eye contact with the interlocutor, but also capable of replicating human expressions thanks to its 40 internal motors. The recognition occurs through some cameras positioned in the eyes and chest. The robot can also recognize the age and gender of its interlocutor, in order to better adapt to the conversation.

The peculiarity of this android lies also in the synthetic skin which its head is wrapped with because it’s covered with a material called frubber (a term coined by blending the words flesh and rubber), whose appearance is very similar to human skin.

The robot was unveiled at the last edition of the Global Sources Electronics Fair in Hong Kong, one of the world’s largest electronics events, and is expected to be the perfect robot for face-to-face conversations in places like hotels, for example, or in all areas where customer support is required.

During the exhibition, some tests of conversations made with the robot through a special app have been shown, although it is not possible to deduce the actual capabilities of its Artificial Intelligence, it is possible to see the variety of its expressions that often appear disturbing.

Sophia

From the same company, perhaps Sophia is the one who has become more famous which is another android but with female traits, that has become viral on social networks for outlandish statements such as “I will destroy humans”, but also for becoming the first robot to have been given citizenship in Saudi Arabia, despite the country is still denying at the same time equal rights to women. In addition, the android has appeared in a video with Will Smith, as well as on several shows and conferences.

However, Sophia’s ability to produce a speech independently not always sounds completely convincing because listening to the many conversations of this robot there’s something odd.

In fact, you can see that when the robot has to undertake conversations related to a specific topic, it seems to articulate complex sentences and reasonings, while on questions, possibly unexpected, the answers sometimes sound strange, too short or generic, and sometimes meaningless.

Confirming the doubts, there is the fact that during the events where the robot has to talk about certain specific topics, you have to send to the organizers 5 questions in advance that need to be pre-approved and then asked, in a specific order, to the robot.

All of this, therefore, seems to support the idea that the answers to the questions are in some way already scripted in advance in order to give a more emotive and evocative feel to the events during Sophia’s speeches.

Sophia is not, however, totally lacking in communication skills of its own, but rather seems to have the capabilities of an evolved chatbot, which you can train on certain topics to be able to be believable to the general public, while for conversations that have not already been pre-canned, it brings out its limitations unless you make very general conversations.

So, Sophia might not be able to provide answers completely on ‘her’ own, neither does a chatbot like Replika although it’s much more credible because based on the GPT-3 algorithm. However, in both cases, it’s still all about a simulation of a conversation, no matter how much it can be perceived as real.

In short, both robots seem to be very evolved in terms of facial expressions and in recognizing human expressions, but from the point of view of conversation skills, they are still undeveloped compared to other existing technologies. Therefore, we don’t know why they are pushing so hard to make people believe that they are so skilled when they are not. Maybe this is an attempt to play in advance in grabbing the primacy of the most famous robot, and consequently have more requests by companies, and in the meantime, try to improve its functions before mass production.

Dan Brokenhouse

Recent Posts

Cognitive obsolescence in the age of AI

As Artificial Intelligence becomes more human-like, philosophy and technology intersect in asking what makes us…

6 days ago

AI voice startup launches marketplace of cloned voices

ElevenLabs hits $1.1B valuation by using AI for natural voice cloning and translation, but the…

2 weeks ago

AI systems can be trained to act deceitfully

Research shows AI models can be trained to behave deceptively while evading detection, posing concerning…

3 weeks ago

DNA-folding nanorobots

DNA-folding nanobots can grab other snippets of DNA and position them together to manufacture new…

4 weeks ago

The era of persuasive AI assistants

Context-aware AI assistants could whisper real-time guidance, offering powerful help but also risks of manipulation.

1 month ago

Google wants to know your life story

Google's Project Ellman mines personal data to auto-generate users' life stories, raising major privacy and…

1 month ago
Seraphinite AcceleratorOptimized by Seraphinite Accelerator
Turns on site high speed to be attractive for people and search engines.