Monthly Archives: July 2023

Excerpted from The Datapreneurs courtesy of Peakpoint Press, an imprint of Skyhorse Publishing.

When artificial general intelligence (AGI), the ability of machines to match or even surpass human intelligence, seems to be within our grasp, we must encode ethics models within our future intelligent machines to ensure they contribute to the betterment of humanity and fulfill the new social contract we need.

While growing up in suburban Michigan, I was the nerdy sort of kid who loved math and science and, not surprisingly, became the AV guy who ran the projector at school. I loved tinkering with electronic gadgets. My other great love as a teenager was science fiction, particularly the work of Isaac Asimov. I read hundreds of his novels and short stories during my teens and twenties. It was not until recently that I began thinking more deeply about intelligent machines ethical and moral implications. I looked back at what Asimov had written about the relationship between people and robots. In addition to featuring robots prominently in his novels and stories, he wrote dozens of essays about them. Asimov concluded they should be respected, not feared or controlled. He developed rules for how they should interact with humanity. Asimov was an optimist and a realist.

I am an optimist and a realist, too. To achieve a future where artificial intelligence benefits humanity, we need a new social contract that governs the relationship between people and the emerging generation of AGI machines.

The concept of a social contract emerged in the Age of Enlightenment, the period of rigorous scientific, political, and philosophical questing that spanned the 18th century in Europe. Traditionally, a social contract is an implicit or explicit deal between the government and the people in a country where individuals surrender some of their freedoms and follow the rules laid out by governments in exchange for other benefits and maintenance of the social order. Because social orders are under constant stress, the values, laws, and regulations that embody social contracts require reexamination and modification when new factors come into play.

The new social contract I have in mind would govern the relationship between people and smart machines assuring that people are safe and the newly emerging AGI entities align with our values and interests. It would consist of a set of rules agreed to by the worlds governments, businesses, and other institutions defining what intelligent machines can and cannot do and how people can and cannot use them.

When I look ahead, I see amazing things coming. We are at the beginning of an intelligence revolution similar to our world in the 1850s during the Industrial Revolution.

This computer science journey began in earnest in the 1940s and 1950s. It was not until the 1960s that universities began offering computer science degrees. Then, the PC, the web, the smartphone, big data, the cloud, foundation models, and tremendous advances in artificial intelligence came in rapid succession. The science and industry that emerged from the IT revolution are still changing rapidly but are also maturing. Technology transforms our world, businesses, and our personal lives.

Now we are on the verge of another great lift. By harnessing AI and other techniques to master the explosion of data we see today, we can understand how the world works much more accurately and comprehensively. We can make better decisions and use Earths resources more responsibly. Throughout my career, I have focused on bringing technology to bear to help businesses succeed. As computer technology developed, its potential impact for good or potential harm increased. I am hopeful these coming technological advances will improve the well-being of our species and the sustainability of life on this planet.

Over the past 80 years of the computing revolution, intelligent machines matched or bested one human capability after another. First, we created machines with expertise in a single domain. With the arrival of foundation models, the depth of AI systems knowledge, the speed with which they react or predict, and the accuracy of their predictions are pretty darned impressive. They already demonstrate a recall of knowledge far beyond human capabilities. Now, we create machines with expertise in multiple domains. These large-scale machine learning models will dramatically lower the cost of intelligence, enabling new smarts and capabilities in applications and services of all types.

I believe machines will possess artificial general intelligence within the next decade.

I see the 2030s and beyond as the golden era of robotics. Todays robots that make cars on assembly lines and clean up nuclear power plants after meltdowns are impressive. But the robots of the future will impress on another order of magnitude. They will be largely autonomous because they cannot always take orders from computing systems located in the cloud. To make machines capable of autonomy, we will have to provide them with AI capabilities that are miniaturized and localized. Initially, these machines will serve single purposes like cleaning our floors, delivering packages, driving vehicles, and flying us around. Over time, more general purpose robots that take on human characteristics and form will emerge.

I believe machines will possess artificial general intelligence within the next decade. It is only a matter of when. These prospects do not frighten me, but they do concern me. What is the societal impact in a world where smart machines are general purpose, matching the capabilities of people and exceeding them in many ways? What ethics and rules will control these machines? In the future, it seems likely that robots will be capable of performing most physical tasks, and intelligent models within them will be capable of performing most intellectual tasks. What will people do if machines and AI systems do all that work?

I am not an expert in AI, and I am not an ethicist. I am an engineer and a businessman. I do not have a clear answer to these issues. These questions will likely be among societys most critical policy issues in the decades ahead. Computer scientists, business leaders, government officials, academics, ethicists, and theologians must work together.

I believe people will develop solutions to the profound ethical issues raised by tomorrows robots and intelligent machines, but I think the process will be messy. In history, every major technological advance has been used, for good and bad. Ultimately, though, common sense prevails, and society establishes laws and regulations that oversee the use of technology. This governance applies to everything from electricity to nuclear technology, and I believe the same will happen with intelligent machines.

We can and will overcome these challenges, and the rising tide can lift all boats. But these issues will not solve themselves. We must think deeply about them and design solutions before the disruptions take full force.

Visit link:

We need a new social contract for the coming golden age of robotics - Big Think

The global transition to renewable energy sources is rapidly gaining momentum, with solar power playing a crucial role in the shift towards a more sustainable future. As the adoption of solar panels continues to grow, the need for efficient and cost-effective maintenance of these installations has become increasingly important. One of the most promising solutions to this challenge lies at the intersection of autonomous robotics and solar panel cleaning technologies.

Solar panels are highly dependent on their ability to absorb sunlight, and even a thin layer of dust or dirt can significantly reduce their efficiency. According to the National Renewable Energy Laboratory, dirty solar panels can lose up to 25% of their energy output, which translates to a considerable loss in potential revenue for solar farm operators. To maintain optimal performance, solar panels need to be cleaned regularly, which can be a labor-intensive and costly process, especially for large-scale installations.

This is where autonomous robotics come into play. By combining advanced robotics with innovative cleaning technologies, it is now possible to create automated systems that can efficiently clean solar panels without the need for human intervention. These robotic solutions not only save time and labor costs but also minimize the risk of damage to the panels and reduce water consumption, making them an environmentally friendly option.

One of the key factors driving the synergy between autonomous robotics and solar panel cleaning technologies is the rapid advancement in artificial intelligence (AI) and machine learning. These technologies enable robots to navigate complex environments, recognize patterns, and make decisions based on real-time data. In the context of solar panel cleaning, AI-powered robots can analyze the level of dirt on the panels, determine the most efficient cleaning method, and adapt their movements to the specific layout of the solar farm.

Another important aspect of this synergy is the development of specialized cleaning tools and techniques designed specifically for solar panels. Traditional cleaning methods, such as pressure washing or manual scrubbing, can cause damage to the delicate solar cells and are not always effective in removing stubborn dirt and grime. Robotic cleaning systems, on the other hand, can be equipped with advanced technologies such as electrostatic cleaning, ultrasonic cleaning, or even laser cleaning, which can remove dirt and dust without causing any harm to the panels.

The integration of autonomous robotics and solar panel cleaning technologies is already showing promising results in the field. Companies like Ecoppia, a leading provider of robotic solar panel cleaning solutions, have developed fully autonomous systems that can clean large-scale solar installations with minimal water usage and without the need for human intervention. These systems have been deployed in solar farms around the world, demonstrating their effectiveness in maintaining the performance of solar panels and reducing operational costs.

As the demand for clean energy continues to grow, the need for efficient and sustainable maintenance solutions for solar installations will become even more critical. The intersection of autonomous robotics and solar panel cleaning technologies offers a promising path forward, enabling solar farm operators to maximize the performance of their assets while minimizing environmental impact and operational costs.

In conclusion, the synergy between autonomous robotics and solar panel cleaning technologies is poised to revolutionize the way solar installations are maintained, ensuring that they continue to deliver clean, renewable energy for years to come. As AI and machine learning continue to advance, we can expect even more sophisticated robotic solutions to emerge, further enhancing the efficiency and sustainability of solar power generation.

View original post here:

The Intersection of Autonomous Robotics and Solar Panel Cleaning ... - EnergyPortal.eu

In a significant advancement in robotics, Google DeepMind has introduced a new AI agent named RoboCat. This agent is designed to learn a variety of tasks across different robotic arms, showcasing the ability to self-generate new training data to improve its techniques, marking a crucial step towards the creation of general-purpose robots.

RoboCat, a Transformer model with a VQ-GAN encoder, was released in June 2023. It is primarily intended for research into learning to accomplish a wide variety of tasks from expert demonstrations or multiple real robot embodiments for manipulation.

The primary intended users are Google DeepMind researchers, and it's not intended for commercial or production use.

RoboCat's standout feature is its learning speed. It can master a new task with as few as 100 demonstrations, leveraging a large and diverse dataset. This capability reduces the need for human-supervised training, potentially accelerating the pace of robotics research.

RoboCat's training involves a comprehensive five-step self-improvement process. It starts with collecting 100-1000 demonstrations of a new task or robot, using a robotic arm controlled by a human. This new task or arm data is used to fine-tune RoboCat, creating a specialized spin-off agent. This agent practices the new task or arm an average of 10,000 times, generating more training data.

The demonstration data and self-generated data are then incorporated into RoboCats existing training dataset, and a new version of RoboCat is trained on the updated dataset.

This process enables RoboCat to learn from a wide range of tasks and diverse training data types. Having been trained on millions of trajectories from both real and simulated robotic arms, RoboCat handles a variety of tasks involving different objects and variations, sourced from Reinforcement Learning (RL), Teleoperation (Teleop), and RoboCat itself.

These tasks include stacking RGB objects, tower and pyramid building with RGB objects, and lifting NIST-i gears, among others. The training involved four different types of robots and many robotic arms to collect vision-based data representing the tasks RoboCat would be trained to perform.

RoboCat demonstrates impressive adaptability by quickly learning to operate different robotic arms. For example, after observing 1000 demonstrations controlled by humans, RoboCat could successfully direct a new arm with a three-fingered gripper and twice as many controllable inputs, achieving an 86% success rate in picking up gears.

Moreover, the more new tasks RoboCat learns, the better it gets at learning additional new tasks. The initial version of RoboCat achieved a 36% success rate on previously unseen tasks after learning from 500 demonstrations per task. However, the latest version, trained on a more diverse set of tasks, more than doubled this success rate on the same tasks.

RoboCat's performance was evaluated through various tasks, such as inserting and removing objects from a bowl and lifting large gears. These evaluations were conducted in both simulated and real-world environments and compared to the performance of human teleoperators.

During the training process, RoboCat uses different observations to understand the robot's position and grip. These observations include joint angles, TCP position, gripper joint angle, and gripper grasp status. The specific observations depend on the robot and objects being used.

In the development of RoboCat, an interesting comparison was made between the VQ-GAN tokenizer and the patch ResNet used in Gato. The patch ResNet tokenizer performed better during training tasks but performed worse on tasks that were not included during training.

It's important to note that RoboCat is currently an early research model and has not been evaluated for deployment and safety outside of research environments. As RoboCat's capabilities expand, potential ethical and safety risks need to be carefully addressed. Therefore, caution should be exercised when considering the use of RoboCat outside of research settings. Nonetheless, the development of RoboCat represents a significant milestone in the field of robotics and AI, bringing us closer to a future where robots are an integral part of our everyday lives.

Read more:

RoboCat: Google DeepMind's innovative leap into AI-powered ... - The Jerusalem Post

ChatGPT and Epic Systems

This years buzzword in academic and now business technology is ChatGPT. Epic Systems, Microsoft, and OpenAI (the owners of ChatGPT) have announced the release of the next evolution in artificial intelligence (AI) called GPT-4. This new system integrates AI with electronic health records with the hopes of advancing medical care and associated services.

In-basket messaging is the first section of electronic health records that will be significantly changed. Typically, provider emails, phone calls, and patient messages are loaded in chronological order. With the integration and advancement of AI, in-basket messages are prioritized and distributed to make the most efficient use of provider time.

Presenters at HIMSS 2023 lectured about the potential application of language models like GPT-4 for automating clinical documentation, decision support, and prescribing. This advancement is done by software that has been trained using a language model on a large dataset of health care-related messages and responses. Eventually, it may be possible to develop a system that can generate appropriate responses to different types of in-basket messages automatically.

However, it is important to note that the use of such systems would need to be carefully monitored and controlled to ensure that patient privacy and confidentiality are maintained at all times. Additionally, a health care provider must carefully review any automated response generated by a language model to ensure that it is accurate and appropriate for the specific context.

Mixed reality (medical extended reality)

Medical extended reality (MXR) or the use of virtual, augmented, and mixed reality technologies in health care is a relatively new field that is being explored for its potential to improve patient care and medical training. Exhibits at the conference demonstrated a variety of ways that MXR can be used, such as medical training, patient education, pain management, and patient rehabilitation.

In the realm of medical training, MXR can be used to simulate medical scenarios and procedures for training purposes. Medical students, nurses, and training allied professionals can use virtual and augmented reality to practice surgical procedures, diagnose medical conditions, or learn about human anatomy.

Medical extended reality can also be used to educate patients about their medical conditions and treatment options. For example, virtual reality can be used to show patients what happens during a surgical procedure or how their medications work in their bodies. It is hoped that using this advanced technology can help in some of the most pressing aspects of modern medical care, such as pain management. For instance, it can be used to distract patients from pain and discomfort post-medical procedures, similar to using a mirror box to relieve a patients pain after a limb amputation.

Lastly, MXR has the potential to aid in the rehabilitation of patients. Virtual and augmented reality can be used to simulate real-world scenarios and activities to help patients recover from injuries and improve their motor skills. There are a number of potential ways to improve patient outcomes and enhance medical education and training. However, it is important to note that MXR is still in the early stages of development, and more research is needed to fully understand its potential benefits and limitations.

Robotics (Cobot)

ChristianaCare has integrated cobots, short for collaborative robots; it is a type of robot designed to assist their medical staff. Cobots complete various tasks throughout the hospital, including reducing the need for human workers to move heavy or bulky items. This can help reduce the risk of workplace injuries and improve efficiency. Cobots are also being used at pharmacies to automate tasks such as medication dispensing and inventory management. It is theorized that this type of automation can help reduce errors and improve efficiency. Overall, cobots are not replacements for human workers but rather a tool to augment their capabilities and improve their working conditions.

Fast Healthcare Interoperability Resources (FHIR)

Fast Healthcare Interoperability Resources intend to deliver the exchange of health care information between different EHR systems. This product helps existing EHR software come into compliance with the EHR interoperability requirements created by the 21st Century Cures Act. Unfortunately, there needs to be much improvement for FHIR to be more meaningful.

Another exciting advancement announced at the HIMSS 23 conference was the partnership between Oracle Health and Athena Health to explore the potential use of biometrics (face recognition directory) to verify the identity of providers. This could be used to increase efficiency for prescribing controlled substances. Many of these exciting developments are still in the planning stages and may not come to fruition.

Afua Aningis a physician informaticist.

See original here:

Seen at HIMSS: the latest innovations in artificial intelligence, mixed ... - Kevin MD