What AI did next

However you choose to play it, it can’t be denied that AI is advancing at an incredible rate across industries as diverse as saving the lives of earthquake survivors, designing spacecraft parts, and developing drugs that can treat opioid addiction. 

To put this into context, James Coop, co-founder of JedAI Studio, says: “AI’s growth and development are now outpacing Moore’s Law exponentially, as in by 50-60 times. In fact, recent research has found that the performance of AI systems over the course of the last decade has doubled every six months or so. This is thanks to a raft of new algorithmic techniques, the availability of large datasets, and increased computational power. 

“But what this means in real-world terms is that we’re on the cusp of an incredible wave of innovation. In short, it’s an incredibly exciting time to be involved in the world of AI.”

And to back up his theory, we’ve put together a list of the top three AI advancements to hit the market this year. But rest assured, not only have there been plenty more than this so far… there will be a whole lot more to come. AI can promise you that. 

–

Nvidia's novel 3D printing tech can produce 3D models from texts

Nvidia, a US-based GPU manufacturer, recently unveiled AI technology called "Magic3D" that could generate 3D models from a text prompt.

A brand-new text-to-3D content generation tool called Magic3D produces 3D mesh models with unmatched quality. They provide customers with new ways to control 3D synthesis, image conditioning techniques, and prompt-based editing methods, opening up new vistas for numerous creative applications, explained the Magic3D developers.

In their paper published on November 18, 2022, the researchers described how this technique would make it possible for anybody to produce 3D models without requiring specialized expertise.

“Once refined, the resulting technology could speed up video game (and VR) development and perhaps eventually find applications in special effects for film and TV. We hope with Magic3D. We can democratize 3D synthesis and open up everyone’s creativity in 3D content creation,” they said.

What jobs can Magic3D handle?

Like DreamFusion, which uses a text-to-image model to create a 2D image that is then optimized into volumetric NeRF (Neural radiance field) data, Magic3D uses a two-stage process that takes a crude model made in low resolution and optimizes it to a higher resolution. According to the paper's authors, the Magic3D method can create 3D objects twice as quickly as DreamFusion.

They use a two-stage coarse-to-fine optimization system to produce text-to-3D material that is both quick and high-quality. Before making a coarse model, they use a low-resolution diffusion in the first step.

They then accelerate using a hash grid and sparse acceleration structure. In the second stage, they employ a textured mesh model that is initialized from the coarse neural representation to enable optimization using a high-resolution latent diffusion model in conjunction with an effective differentiable renderer.

--

Scientists use AI to develop drugs that can treat opioid addiction

Leslie Salas Estrada, in the lab of Marta Filizola, at the Icahn School of Medicine at Mount Sinai, now hopes to alleviate opioid addiction by discovering drugs that inhibit the kappa-opioid receptor. 

“If you’re addicted, and you’re trying to quit, at some point, you will get withdrawal symptoms, and those can be really hard to overcome,” Estrada explained to NeuroscienceNews.

“After a lot of opioid exposure, your brain gets rewired to need more drugs. Blocking the activity of the kappa opioid receptor has been shown in animal models to reduce this need to use drugs in the withdrawal period.”

The challenge of this task is in uncovering the drugs that can actually block the activity of a protein, such as the kappa-opioid receptor, in a sea of countless candidates. That’s why Estrada turned to computational tools to make the process more efficient. Estrada is using artificial intelligence (AI) to optimize her drug-finding systems.

“Artificial intelligence has the advantage of being able to take huge amounts of information and learn to recognize patterns from it. So, we believe that machine learning can help us to leverage the information that can be derived from large chemical databases to design new drugs from scratch. And in that way, we can potentially reduce the time and costs associated with drug discovery,” she said.

Estrada’s team trained a computer model to generate compounds that might block the receptor with a reinforcement learning algorithm that rewarded properties that are favorable for drug treatments. They did this by combining information about the kappa-opioid receptor and known drugs.

It proved successful thus far. The researchers have already identified several compounds that have promising attributes. They are now aiming to synthesize them and eventually test them in animal models for safety and effectiveness. 

The ultimate goal, Estrada said, is to” help people struggling with addiction,” according to NeuroscieNews.

--

Balancing act: AI-powered superhuman robo-boots coming soon

Researchers at Georgia Tech and Emory used a motorized floor to literally “pull the rug out” from under healthy young participants – pitching them forward toward the ground. We asked them to stay balanced with their feet in place, but sometimes the slip was so large that they had to take a step to avoid face-planting on the ground. Then, we programmed the exo-boots to give the participants a blast of assistance with either the same delay as their natural response or artificially faster than humanly possible and compared that to giving no exo-boot assistance at all.

The researchers say that they were surprised to find that only the mode that beats the human reaction to the punch helped users recover balance faster and prevented them from taking a step to recover.

At the same time, they recorded physiological responses of our exo-boot “test pilots” to see whether the device was mimicking – or potentially interfering with – their underlying balance responses. Using ultrasound, we could look under the skin to see how calf muscles stretched during the slip. Muscle stretch generates critical sensory signals needed by the nervous system to initiate a balance reaction.

The faster-than-human exo-boot balance response actually eliminated calf muscle stretch signals, but the control signals to those same muscles, in response to the nervous system generally, persisted. This finding highlights that the nervous system is more than a set of simple reflexes that react to local muscle stretch but instead acts to gather information from throughout the body to remain upright in both standing and walking.

Why it matters

People tend to take balance for granted. But every step holds the possibility of a fall, and bad falls are a top reason for health declines when people grow older. If researchers can develop assistive technology to improve balance while getting from here to there, it can prevent falls and enable people to be active for longer.

What other related research is going on in the field

Using wearable robots to help mobility, particularly balance, has long been a dream of scientists and engineers. But research to date has focused primarily on providing powered assistance to make it easier to walk faster or longer. Aside from a handful of promising laboratory-based research devices, the field of wearable robots has not focused much on balance.

Technology advances in the past decade open up new possibilities for state-of-the-art wearable robots to improve balance. Other researchers around the world are also adapting existing exoskeleton systems to assist balance using biologically inspired control algorithms that may, in time, be able to respond faster than a human’s reflex response.

What’s next

The study is a proof-of-concept demonstration that exoskeletons can improve balance in a controlled laboratory setting during a simple upright standing task for young adults without balance problems. There is much to be done to enable wearable robots to help with balance in everyday life and improve life for people with balance impairments like older adults, or people with stroke or spinal cord injury. Future studies will be required to explore using exoskeletons at other lower-limb joints, like the knee and hip, to develop artificial intelligence that can anticipate the risk of falling and enable personalized superhuman balance.