[ad_1]
I recently watched M3GAN — a sci-fi flick about a sophisticated AI doll designed to be a caretaker and companion for children that goes rogue. While I didn’t find the movie especially well done or fun, I was rather piqued by its premise. So I decided to dig deeper and find out just how close we are to making something like M3GAN a reality.
M3GAN is a humanoid robot built to resemble a tween girl. Despite a petite frame, she is capable of extreme feats of strength and agility thanks to her advanced actuators, joints (motors responsible for robot’s motion) and robust metallic skeleton. In the movie, M3GAN can dance, overcome and restrain a healthy adult woman, lift and hold a fully grown man, run fast — and execute humans and animals in all sorts of gory ways.
A robot capable of such feats should have a robust kinematic structure, superior structural integrity and a refined set of sensors and snappy actuators to enable fluid movement. More importantly, it needs a powerful processor to take in all this sensory data and process it in real time.
M3GAN is most likely powered by an energy source with extremely high density necessary to retain small form factor while still providing enough to power robotic functions. When she is not causing havoc, we see her reclined on a bench in a child’s bedroom, sitting on what looks like a wireless charging point. It most likely takes her an entire night to recharge for another day of mayhem and murder.
How does modern robotics compare? The most agile robots currently available are produced by Boston Dynamics, a unit of Hyundai Motor Co.
HYMTF,
Maybe you’ve seen its Spot robot dance and Atlas’ parkour prowess. Although some of the movements are still undeniably rigid, this level of agility would still be quite sufficient to roughly mimic what M3GAN can do.
We don’t know how much M3GAN weighs, although in one scene a teenage boy effortlessly carries her around and throws her to the ground, implying a rather lightweight chassis. Her battery life also isn’t specified, but judging by several scenes in the movie, it’s most likely between eight- and 12 hours. On the other hand, Boston Dynamic’s Atlas weighs 345 pounds and can pull off just one hour of autonomy with its 3.7kw battery. While, in this regard, Atlas is no match for M3GAN, another robot could be — Tesla’s
TSLA,
Optimus has a 2.3Kw battery pack capable of eight-hour autonomy.
We’ve discussed the brawn, let’s now discuss the brains. M3GAN is run by artificial general intelligence (AGI), an AI system with cognitive abilities that should be able to match or exceed human intelligence, capable of performing any intellectual task a human can.
Relying on generative machine learning algorthms, M3GAN, or Model 3 Generative Android, can not only come up with new ideas and concepts, but can also, thanks to AGI, learn like humans. In the movie, she is designed to be the best child companion, and her AI is trained on dataset of millions of hours of recorded interaction between children and their toys.
While AGI exists in the world of M3GAN, in real world we’re not quite there — yet.
Most AI researchers have not focused on AGI, with some believing it’s too complicated to replicate in the near future. However, a few computer scientists are actively involved in AGI research and contribute to AGI conferences, producing diverse and groundbreaking research.
One of the most popular examples of what could be classified as the first step toward AGI is DeepMind’s Gato — a “general-purpose” system capable of performing more than 600 different tasks.
Now that we’ve covered both the brains and the power behind M3GAN and their closest counterparts in the real world, one question remains: Which of these two companies is closest to creating M3GAN in regard to its purpose and the risks it poses?
Atlas is designed for flashy display of power and speed, while Optimus focuses on efficiency. Watching the Tesla Optimus walk may seem laughable compared to Boston Dynamic’s agile model, but bear in mind that Optimus is geared for factory work and civilian use. Atlas is currently too bulky and power-hungry to be used for anything but tasks requiring short bursts of strength and functionality. Boston Dynamics has perfected kinetics and kinematics necessary for fluid movement, but to make Atlas more practical, the company has yet to master complex AI and autonomy-based optimization.
Optimus was built on Tesla’s robust AI model — the same one used for full self-driving in its vehicles. Optimus uses this data not only to navigate obstacles, but also to grab and manipulate objects.
While Atlas was designed to be used in search and rescue operations, Tesla’s model is a product built around economic principles: it needs to provide value to the customer while being profitable to the manufacturer. Tesla will need to be conscious of its cost while efficiently solving engineering and AI problems at hand. It also means that, in time, Tesla will further improve the AI and the hardware of Optimus, turning the robot into an indispensable tool on the conveyor belt, as well as a home companion.
Needless to say, there’s a long road ahead for both of these models, but it’s more likely that Optimus will win that race.
Would you have a robot companion in your household? Let me know in the comment section below.
More: ChatGPT is challenging Google’s control over internet search — setting up an epic clash of AI titans
Also read: Bitcoin and crypto are moving past FTX’s collapse, but the road back will be rough
[ad_2]
Source link
