Towards AI Can Help your Team Adopt AI: Corporate Training, Consulting, and Talent Solutions.


Realbotix and MelNet… Because they Already Can
Latest   Machine Learning

Realbotix and MelNet… Because they Already Can

Last Updated on July 20, 2023 by Editorial Team

Author(s): Dr. Adam Hart

Originally published on Towards AI.

In the 15th Century, mannequins, a type of life-size doll with static articulated limbs, evolved to support the work of dressmakers.

At the turn of the 20th Century, David Hilbert famously outlined his 23 unsolved problems for mathematics, later declaring, “We must know, we will know.”. [1]

In the 21st Century, we have a curious conjunction of mannequin and math: Realbotix. They have a companion product Real Doll X. It appears their product is geared to a male clientele at this time.

You can see a video of their mission statement and a doll here.

Their rationale is founded on similar claims that many AI and robotic efforts purport; angled towards solving human problems or of illness or perhaps more accurately understood in this context, a purported human limitation of loneliness.

Curiously, a 2018 fictional Russian Netflix series named Better Than Us has a similar premise.

Realbotix, in case you are unaware, is a conjunction of long-running trends in mathematics and an evolutionary human preoccupation with creating artificial forms of itself in many guises. Excepting that in this case, enabled by technological advances in speech recognition, robotics, pattern analysis and materials, they can mimic a speech-response and supposedly fake “empathy”.

Perhaps similar to the way 9th Dan Go master Lee Sedol marveled at AlphaGo’s “creativity” per Curious’s recent opinion post regarding DeepMind’s AlphaGo.

Now, it is a facet of technological evolution that often times inventions or changes are created in the absence of a need, a problem, or even a market.

So why would they do this, create something that even twenty years ago would not be acceptable to society, and why now?

The simple answer is curiosity. Not unlike a child playing with matches to experiment, not unlike the Feynman picking the lock on the filing cabinet in the Manhattan project as a prank, to learn, to push the boundaries of human limitation, to combine existing inventions in novel ways that produce a new thing. Just like the original iPhone was not new tech, instead was a new combination of existing tech.

These are intelligent and motivated adult scientists, engineers and artists; funded by IPO and private equity raising, and they have made the world’s first companion sex doll that supposedly mimics empathy.

The more complex answer is because they already can.

Very recently, Facebook engineers revealed a superior algorithm for mimicry of the human voice. They called it MelNet.

MelNet got press for its uncanny ability to mimic famous people. It is a type of generative neural network that was trained using TED talk videos. And it can fool humans-at-a-distance, which means if Bill Gates is not in front of you, and you’ve heard him speak before, it’s unlikely you can tell the difference.

In the Realbotix video link above, the inventor calmly talks about the possibility of the acceleration of AI capability that in the next 10–15 years may mean that AI can achieve consciousness. “We’ll have to wait and see.”

What is a consistent trend with scientific advancement and especially at this juncture in history concerning AI, is that many reports surface in the press that could be disturbing, disconcerting or downright worrying, but we don’t yet see it in our day to day lives, or we disbelieve it; or, we are awestruck and horrified because we don’t understand what is happening to us, at least not until much later.

In the same way in the last decade or so smartphones have become ubiquitous, AI will too. There is too much critical mass. And it will evolve in unusual and odd directions, like Realbotix, because of human curiosity, and because it can.

It may already be the case that Realbotix is seeking to take advantage of the published MelNet advancement to offer the opportunity to replace their voice with a voice of a deceased loved one, a former partner, a famous personage, or yourself.

Who has intellectual property rights over a recording of your voice? Once the recording has been captured, by speaking with Siri, Google Home, Amazon Alexa, the digitized version of your voice is captured, cataloged, identified by your fixed IP address which is inextricably linked with your ISP. Governed legally by the terms of service agreement that we mostly scroll to the bottom of and tick a box and hit “next”.

In the aforementioned 2018 fictional series Better Than Us series, an astute observer notices the use of vintage cars like 1980 Volvos coexisting with the hi-tech robots that can perform miracles like inoperable brain surgery.

It is the highly advanced technology that today coexists with the familiar technologies like electric toasters and kettles, layered and interrelated with our day-to-day lives. The human ability to adapt schemers them together in one coexisting experience.

One palpable tradeoff we are making in this coexistence is convenience and marvel versus privacy and exposure to fakery in our homes. But it wasn’t us that invented this relationship; it was technology-looking-for-a-need that did this.

It is still possible to make toast over an open flame, but the bread is apt to burn. The thermostat in the toaster regulates the temperature.

The difference between a toaster and a Real Doll X perhaps voiced by MelNet is that the toaster is not intending to mimic or fake a human or play inauthentically on our emotions, that very facet which makes us human.

Arguments for the regulation of technological advancement fly in the face of human curiosity and ingenuity, and technologies are historically often coopted later for purposes other than what the inventor intended. Once it is in the wild it is there to be reused and repurposed. The ubiquity of a good solution is achieved through meeting a genuine need in an efficient, cost effective and human understandable and controllable manner.

Witness the Edison light bulb. Prior to that time, gaslighting sometimes burned houses and many theatres down, or caused asphyxiation.

These latest fragments emerging from the AI agenda will need to meet the same criteria for survival as well. Why efforts by Sand Hill Road funded mathematicians, engineers and scholars are being placed into these odd pathways instead of others like sustainable food supply, sustainable population growth distribution, carbon sequestration, avoidance of desert biomes is, because of they, Hilbert’s people, can. And they patently can already.


[1] an inversion of the Latin Ignoramus et ignorabimus. Which infact Demis Hassabis of DeepMind fame has recently invoked to highlight his AI achievements, as helping overcome limitations in human knowledge.

Originally published at

Join thousands of data leaders on the AI newsletter. Join over 80,000 subscribers and keep up to date with the latest developments in AI. From research to projects and ideas. If you are building an AI startup, an AI-related product, or a service, we invite you to consider becoming a sponsor.

Published via Towards AI

Feedback ↓