AI has finally come full circle.

A new suite of algorithms by Google Brain can now design computer chips—those specifically tailored for running AI software—that vastly outperform those designed by human experts. And the system works in just a few hours, dramatically slashing the weeks- or months-long process that normally gums up digital innovation.

At the heart of these robotic chip designers is a type of machine learning called deep reinforcement learning. This family of algorithms, loosely based on the human brain’s workings, has triumphed over its biological neural inspirations in games such as Chess, Go, and nearly the entire Atari catalog.

But game play was just these AI agents’ kindergarten training. More recently, they’ve grown to tackle new drugs for Covid-19, solve one of biology’s grandest challenges, and reveal secrets of the human brain.

In the new study, by crafting the hardware that allows it to run more efficiently, deep reinforcement learning is flexing its muscles in the real world once again. The team cleverly adopted elements of game play into the chip design challenge, resulting in conceptions that were utterly “strange and alien” to human designers, but nevertheless worked beautifully.

It’s not just theory. A number of the AI’s chip design elements were incorporated into Google’s tensor processing unit (TPU), the company’s AI accelerator chip, which was designed to help AI algorithms run more quickly and efficiently.

“That was our vision with this work,” said study author Anna Goldie. “Now that machine learning has become so capable, that’s all thanks to advancements in hardware and systems, can we use AI to design better systems to run the AI algorithms of the future?”

The Science and Art of Chip Design

I don’t generally think about the microchips in my phone, laptop, and a gazillion other devices spread across my home. But they’re the bedrock—the hardware “brain”—that controls these beloved devices.

Often no larger than a fingernail, microchips are exquisite feats of engineering that pack tens of millions of components to optimize computations. In everyday terms, a badly-designed chip means slow loading times and the spinning wheel of death—something no one wants.

The crux of chip design is a process called “floorplanning,” said Dr. Andrew Kahng, at the University of California, San Diego, who was not involved in this study. Similar to arranging your furniture after moving into a new space, chip floorplanning involves shifting the location of different memory and logic components on a chip so as to optimize processing speed and power efficiency.

It’s a horribly difficult task. Each chip contains millions of logic gates, which are used for computation. Scattered alongside these are thousands of memory blocks, called macro blocks, which save data. These two main components are then interlinked through tens of miles of wiring so the chip performs as optimally as possible—in terms of speed, heat generation, and energy consumption.

“Given this staggering complexity, the chip-design process itself is another miracle—in which the efforts of engineers, aided by specialized software tools, keep the complexity in check,” explained Kahng. Often, floorplanning takes weeks or even months of painstaking trial and error by human experts.

Yet even with six decades of study, the process is still a mixture of science and art. “So far, the floorplanning task, in particular, has defied all attempts at automation,” said Kahng. One estimate shows that the number of different configurations for just the placement of “memory” macro blocks is about 10^2,500—magnitudes larger than the number of stars in the universe.

Game Play to the Rescue

Given this complexity, it seems crazy to try automating the process. But Google Brain did just that, with a clever twist.

If you think of macro blocks and other components as chess pieces, then chip design becomes a sort of game, similar to those previously mastered by deep reinforcement learning. The agent’s task is to sequentially place macro blocks, one by one, onto a chip in an optimized manner to win the game. Of course, any naïve AI agent would struggle. As background learning, the team trained their agent with over 10,000 chip floorplans. With that library of knowledge, the agent could then explore various alternatives.

During the design, it worked with a type of “trial-and-error” process that’s similar to how we learn. At any stage of developing the floorplan, the AI agent assesses how it’s doing using a learned strategy, and decides on the most optimal way to move forward—that is, where to place the next component.

“It starts out with a blank canvas, and places each component of the chip, one at a time, onto the canvas. At the very end it gets a score—a reward—based on how well it did,” explained Goldie. The feedback is then used to update the entire artificial neural network, which forms the basis of the AI agent, and get it ready for another go-around.

The score is carefully crafted to follow the constraints of chip design, which aren’t always the same. Each chip is its own game. Some, for example, if deployed in a data center, will need to optimize power consumption. But a chip for self-driving cars should care more about latency so it can rapidly detect any potential dangers.

The Bio-Chip

Using this approach, the team didn’t just find a single chip design solution. Their AI agent was able to adapt and generalize, needing just six extra hours of computation to identify optimized solutions for any specific needs.

“Making our algorithm generalize across these different contexts was a much bigger challenge than just having an algorithm that would work for one specific chip,” said Goldie.

It’s a sort of “one-shot” mode of learning, said Kahng, in that it can produce floorplans “superior to those developed by human experts for existing chips.” A main throughline seemed to be that the AI agent laid down macro blocks in decreasing order of size. But what stood out was just how alien the designs were. The placements were “rounded and organic,” a massive departure from conventional chip designs with angular edges and sharp corners.

Human designers thought “there was no way that this is going to be high quality. They almost didn’t want to evaluate them,” said Goldie.

But the team pushed the project from theory to practice. In January, Google integrated some AI-designed elements into their next-generation AI processors. While specifics are being kept under wraps, the solutions were intriguing enough for millions of copies to be physically manufactured.

The team plans to release its code for the broader community to further optimize—and understand—the machine’s brain for chip design. What seems like magic today could provide insights into even better floorplan designs, extending the gradually-slowing (or dying) Moore’s Law to further bolster our computational hardware. Even tiny improvements in speed or power consumption in computing could make a massive difference.

“We can…expect the semiconductor industry to redouble its interest in replicating the authors’ work, and to pursue a host of similar applications throughout the chip-design process,” said Kahng.

“The level of the impact that [a new generation of chips] can have on the carbon footprint of machine learning, given it’s deployed in all sorts of different data centers, is really valuable. Even one day earlier, it makes a big difference,” said Goldie.

The post A Google AI Designed a Computer Chip as Well as a Human Engineer—But Much Faster appeared first on Artificial Intelligence.

Will computer vision take human jobs in the upcoming years with cutting-edge technologies?

There is a worldwide controversy whether computer vision will take over human jobs that can increase the rate of unemployment. The integration of cutting-edge technologies with AI algorithms into the existing computer system has introduced advanced computer vision. AI algorithms transform multiple sets of real-time data into appropriate business insights without any human intervention. But AI algorithms and computers require human assistance to complete multiple tasks efficiently and effectively. Have you ever wondered why there is an increase in human jobs in the field of AI in these recent years? Human skills are needed for the advancement in software development as well as innovating new technologies to boost productivity. Reputed companies and startups provide job opportunities such as computer vision engineer, computer vision scientist, deep learning specialist, software developer, data scientist, software engineer, lead scientist, data analytics lead, computer vision research engineer, and many more with lucrative salary package. That being said, we can claim that computer vision will not take human jobs but will ease the workload effectively to achieve higher ROI.

• PLANNING ON BECOMING A COMPUTER VISION ENGINEER? HERE’S WHAT YOU NEED TO KNOW
• INTEL’S MOVE TO LEVERAGE COMPUTER VISION SOLUTIONS
•  SIGNIFICANT BREAKTHROUGHS AND COUNTRIES IN COMPUTER VISION TECHNOLOGY
• COMPUTER VISION VS HUMAN VISION: FILLING THE VOID IS INDEED DIFFICULT

How does computer vision ease the workload in human jobs?

Computers analyze multiple sets of raw data from digital images and videos to provide appropriate decisions by understanding the environment through the new form of AI known as computer vision. The data is more accessible and affordable due to digital transformation and globalization. Computer vision is thriving in recent years because of the accuracy rates for image pattern recognition than humans. Deep learning neural networks enable the iterative learning process in computers to acquire, process, and analyze image patterns efficiently and effectively than the human visual cognitive system. Convolutional Neural Network (CNN) is used in computer vision technology for appropriate image pattern recognition. These neural networks scan the available image pixel by pixel to identify patterns and memorize the ideal output from different characteristics such as contours and colors. Humans are needed to develop smart machines for completing automated tasks with visual cognition.

The computer vision system can be used for object classification, object identification, and object tracking. It takes much lesser time to analyze thousands of images as well as to detect any defect or issue with hi-tech cameras, data, and AI algorithms than the naked human eyes. Let’s explore the potential of computer vision across several industries to boost productivity.

• Automotive: Computer vision through ADAS, RADAR as well as LIDAR technologies provide visual representations, high visibility, and 3D representations of the surroundings respectively. Automotive Gesture Recognition also monitors the facial and hand gestures of drivers with audible and visual alerts.
• Retail: Computer vision helps in security through CCTVs, spillage detection, theft control, video analytics, enhancing the shopping experience, optimizing operations, alerting in-shelf productivity, and better customer engagement
• Manufacturing: Computer vision helps factory workers in predictive maintenance, identifying defects and eliminating risks as well as product quality inspection for minimal waste of products
• Healthcare: Computer vision can detect any unusual image pattern in reports and X-rays accurately, early-stage tumors, arteriosclerosis, and many more for doctors and nurses to operate at the right time
• Agriculture: Computer vision detects pests and plant diseases, information about high-quality crops, provides facial recognition to identify the individual animal, and analyses grain quality for farmers efficiently

Thus, we all can confirm that computer vision is not taking away human jobs but there is a high probability that it will generate more human job opportunities in the upcoming years. The aim of computer vision is to collaborate with humans to enhance the workload with more appropriate outcomes without any failure or mistake. We have to remember that humans are the main creator behind all these achievements of computer vision.

The post WILL COMPUTER VISION TAKE HUMAN JOBS? appeared first on Artificial Intelligence.

The fifth annual international summit ‘Machines Can See’ will be held on July 8 in Moscow at the Omega Rooftop and is hosted by VisionLabs.

This event brings together the world’s leading experts in computer vision and machine learning to discuss technology trends and share experience, connecting international AI communities.

Human-centric technologies

This year’s theme is “human-centric technologies” with speakers:

· Dima Damen – Associate Professor in the Department of Computer Science at the University of Bristol

· Dr. Efstratios Gavves – Associate Professor at the University of Amsterdam, Scientific Director of the QUVA Deep Vision Lab, Scientific Director of the POP-AART Lab

· Bernard Ghanem – Associate Professor in the CEMSE division, a theme leader at the Visual Computing Center (VCC), and the Interim Lead of the AI Initiative at KAUST

· Ira Kemelmacher-Shlizerman – Associate Professor of Computer Science at the Allen School, Director of the UW Reality Lab, and an Eng Lead at Google

· Kris M. Kitani – Associate Research Professor and Director of the Computer Vision MS program in the Robotics Institute at Carnegie Mellon University

This event also includes a computer vision competition on gesture recognition that runs until July 5. Winners will be announced at the summit and will share a prize fund of 500 000 rubles.

The post Machines Can See – Computer Vision and Deep Learning Summit appeared first on Artificial Intelligence.

Intelligence Amplification (IA), is designed to complement human intelligence.

Intelligence amplification (IA) or cognitive augmentation or machine augmented intelligence was first proposed in the 1950s and 1960s by cybernetics and early computer pioneers. Intelligent Amplification, a novice term is used to describe the effective use of information technology to augment human intelligence.

IA is often contrasted with AI (artificial intelligence). AI is built on the sole purpose of making the computer’s smarter with human-like intelligence to perform autonomous workflows. To succeed, AI needs better models and data, and IA aims to fill the lacuna.

In contrast to AI, which is a standalone system capable of processing information, IA is designed to complement and amplify human intelligence.

Intelligence Amplification has one big edge over AI! It builds on human intelligence which has evolved over millions of years, while Artificial Intelligence aims to build Intelligence from scratch. Experts view that writing was among the first technologies which might pass as IA. It enabled us to enhance our creativity, understanding, efficiency and, ultimately build on the ultimate intelligence.

Intelligence Amplification: Base of AI

Although IA has been around for many years, it has not been a widely recognised subject. With systems like HoloLens, IA can now be explicitly developed to be the faster ingredient of AI.

IA, many believe, offers a safer way to develop tools and technologies which gain their efficacy from human consciousness rather than building their own intelligence. Tools developed through IA can be used for a variety of purposes such as natural language tools, developing knowledge bases, electronic discovery and image processing tools to name a few.

Intelligence augmentation uses machine learning technologies which are similar to AI, but instead of replacing humans, IA aims to assist them. In this situation, instead of depending solely on machines for business procedures, IA machine learning works in tandem with the human brain. Intelligence Amplification aims to make the workplace more efficient and productive just like AI, however, it does it in partnership with humans to support new discoveries and problem-solving while Artificial Intelligence is a bit different, it seeks to bypass humans altogether. For instance, let’s consider a machine learning algorithm that can process massive amounts of patient data by searching through patient history, family history, data from wearables, previous records and tests to predict illness. This information will be presented to medical staff in a way that supports the doctor, who would use their reasoning to reach a diagnosis. Because the human element (doctor) is still involved, this is Intelligence Amplification. But if the program is restricted to data sorting and reach a diagnosis, then it would be Artificial Intelligence.

Intelligence Amplification: Real-Time Use Cases

Retail

Integrating intelligence augmentation technology into shopping applications can make it easier for Shoppers to find products. for instance, the Walgreens app links to the user shopping card and uses Google Tango’s computer vision – a 3D mapping service – to guide in-store shoppers using a downloaded map. Customers can search item numbers, product categories and product names by voice and command the mapping app to provide them store navigation instructions to the product’s location in the store.

Engineering

Innovative technology has disrupted manufacturing and production. With IA, there is another tool for maintenance and creation of complicated and expensive machines which makes it easier for engineers to make repairs. For example, NGRAIN, a Canadian 3D imaging company uses 3D analytics to scan and detect minor damage to aircraft and uses the information to repair operations and improve maintenance.

Home Decor

IKEA, one of the most well-known home decor companies, created an app with IA technology. It supports both 2D and 3D image recognition and image tracking. With this technology, users can move still photos of furniture and furnishings into an image of a room. It’s also possible to measure the dimensions of the room through the app and allow the users to drop an image of a piece of furniture into the photo. this way, shoppers can see what their Ikea purchases will look like in their home before deciding what to buy.

Military

Intelligence augmentation could be used in high-fidelity visualisation which magnifies situation awareness and improved mission planning. Using 3D models, it would be possible to carry out successful operations in defence.

Medicine

When it comes to complicated medical procedures, using IA technology could improve them significantly. Researchers all over the globe are developing tools. Cambridge consultants have determined potential solutions by combining CT and MRI scans to build 3D images of a patient’s body. This assist surgeons to perform more precise surgical processes.

Summing up, there’s no war between the IA and AI. Both of them play vital roles in our day-to-day lives, as we try to figure out which one is here to stay for the long run. As technology continues to advance, IA can be deployed to help address many of the challenges associated with AI, which is why the two technologies must work cohesively. Doing so would translate to the betterment of society as a whole.

The post HOW DOES INTELLIGENCE AMPLIFICATION MAKE SMARTER AI? appeared first on Artificial Intelligence.

Researchers from Cornell University have created a computer algorithm inspired by the mammalian olfactory system. Scientists have long sought out explanations of how mammals learn and identify smells. The new algorithm provides insight into the workings of the brain, and applying it to a computer chip allows it to quickly and reliably learn patterns better than current machine learning models. 

Thomas Cleland is a professor of psychology and senior author of the study titled “Rapid Learning and Robust Recall in a Neuromorphic Olfactory Circuit,” published in Nature Machine Intelligence on March 16.

“This is a result of over a decade of studying olfactory bulb circuitry in rodents and trying to figure out essentially how it works, with an eye towards things we know animals can do that our machines can’t,” Cleland said. 

“We now know enough to make this work. We’ve built this computational model based on this circuitry, guided heavily by things we know about the biological systems’ connectivity and dynamics,” he continued. “Then we say, if this were so, this would work. And the interesting part is that it does work.”

Intel Computer Chip

Cleland was joined by co-author Nabil Imam, a researcher at Intel, and together they applied the algorithm to an Intel computer chip. The chip is called Loihi, and it is neuromorphic, which means it is inspired by the functions of the brain. The chip has digital circuits that mimic the way in which neurons learn and communicate. 

The Loihi chip relies on parallel cores that communicate via discrete spikes, and each one of these spikes has an effect that can change depending on local activity. This requires different strategies for algorithm design than what is used in existing computer chips. 

Through the use of neuromorphic computer chips, machines could work a thousand times faster than a computer’s central or graphics processing units at identifying patterns and carrying out certain tasks. 

The Loihi research chip can also run certain algorithms while using around a thousand times less power than traditional methods. This is well-suited for the algorithm, which can accept input patterns from various different sensors, learn patterns quickly and sequentially, and identify each of the meaningful patterns even with strong sensory interference. The algorithm is capable of successfully identifying odors, and it can do so when the pattern is an astounding 80% different from the pattern originally learned by the computer. 

“The pattern of the signal has been substantially destroyed,” Cleland said, “and yet the system is able to recover it.”

The Mammalian Brain

The brain of a mammal is able to identify and remember smells extremely well, and there can be thousands of olfactory receptors and complex neural networks working to analyze the patterns associated with odors. One of the things that mammals can do better than artificial intelligence systems is retain what they’ve learned, even after there is new knowledge. In deep learning approaches, the network must be presented with everything at once, since new information can affect or even destroy what the system previously learned. 

“When you learn something, it permanently differentiates neurons,” Cleland said. “When you learn one odor, the interneurons are trained to respond to particular configurations, so you get that segregation at the level of interneurons. So on the machine side, we just enhance that and draw a firm line.”

Cleland spoke about how the team came up with new experimental approaches. 

“When you start studying a biological process that becomes more intricate and complex than you can just simply intuit, you have to discipline your mind with a computer model,” he said. “You can’t fuzz your way through it. And that led us to a number of new experimental approaches and ideas that we wouldn’t have come up with just by eyeballing it.”

The post Researchers Develop Computer Algorithm Inspired by Mammalian Olfactory System appeared first on Artificial Intelligence.

Engineers from Duke University and the Institut de Physique de Nice in France have developed a new method to identify objects using microwaves that improves accuracy while reducing the associated computing time and power requirements.

The system could provide a boost to object identification and speed in fields where both are critical, such as autonomous vehicles, security screening and motion sensing.

The new machine-learning approach cuts out the middleman, skipping the step of creating an image for analysis by a human and instead analyzes the pure data directly. It also jointly determines optimal hardware settings that reveal the most important data while simultaneously discovering what the most important data actually is. In a proof-of-principle study, the setup correctly identified a set of 3D numbers using tens of measurements instead of the hundreds or thousands typically required.

The results appear online on December 6 in the journal Advanced Science and are a collaboration between David R. Smith, the James B. Duke Distinguished Professor of Electrical and Computer Engineering at Duke, and Roarke Horstmeyer, assistant professor of biomedical engineering at Duke.

“Object identification schemes typically take measurements and go to all this trouble to make an image for people to look at and appreciate,” said Horstmeyer. “But that’s inefficient because the computer doesn’t need to ‘look’ at an image at all.”

“This approach circumvents that step and allows the program to capture details that an image-forming process might miss while ignoring other details of the scene that it doesn’t need,” added Aaron Diebold, a research assistant in Smith’s lab. “We’re basically trying to see the object directly from the eyes of the machine.”

In the study, the researchers use a metamaterial antenna that can sculpt a microwave wave front into many different shapes. In this case, the metamaterial is an 8×8 grid of squares, each of which contains electronic structures that allow it to be dynamically tuned to either block or transmit microwaves.

For each measurement, the intelligent sensor selects a handful of squares to let microwaves pass through. This creates a unique microwave pattern, which bounces off the object to be recognized and returns to another similar metamaterial antenna. The sensing antenna also uses a pattern of active squares to add further options to shape the reflected waves. The computer then analyzes the incoming signal and attempts to identify the object.

By repeating this process thousands of times for different variations, the machine learning algorithm eventually discovers which pieces of information are the most important as well as which settings on both the sending and receiving antennas are the best at gathering them.

“The transmitter and receiver act together and are designed together by the machine learning algorithm,” said Mohammadreza Imani, research assistant in Smith’s lab. “They are jointly designed and optimized to capture the features relevant to the task at hand.”

“If you know your task, and you know what sort of scene to expect, you may not need to capture all the information possible,” said Philipp del Hougne, a postdoctoral fellow at the Institut de Physique de Nice. “This co-design of measurement and processing allows us to make use of all the a priori knowledge that we have about the task, scene and measurement constraints to optimize the entire sensing process.”

After training, the machine learning algorithm landed on a small group of settings that could help it separate the data’s wheat from the chaff, cutting down on the number of measurements, time and computational power it needs. Instead of the hundreds or even thousands of measurements typically required by traditional microwave imaging systems, it could see the object in less than 10 measurements.

Whether or not this level of improvement would scale up to more complicated sensing applications is an open question. But the researchers are already trying to use their new concept to optimize hand-motion and gesture recognition for next-generation computer interfaces. There are plenty of other domains where improvements in microwave sensing are needed, and the small size, low cost and easy manufacturability of these types of metamaterials make them promising candidates for future devices.

“Microwaves are ideal for applications like concealed threat detection, identifying objects on the road for driverless cars or monitoring for emergencies in assisted-living facilities,” said del Hougne. “When you think about all of these applications, you need the sensing to be as quick as possible, so we hope our approach will prove useful in making these ideas reliable realities.”

The post Machine Learning Shapes Microwave for a Computer’s Eyes appeared first on Artificial Intelligence.

Artificial intelligence (AI) is a vague, slightly tech-y term that is overused by marketing departments trying to show some bona fides. But if you want some real insights on what artificial intelligence is and what it can do, then you should check out these talks that Chris Satchell of Zume and Erik Andrejko of Evolv (the venture arm of Kraft-Heinz) each did onstage at our recent Smart Kitchen Summit.

But before you can begin talking about AI you must understand the importance of data. Not just plenty of it or the right kind, as Satchell points out in his presentation. Before you can get into algorithms you need clean, tagged data that is centralized. He recommends that you don’t spread out your data teams across the company, and instead cluster them together to get the best results.

With your data organized and your models in place, then you can start to analyze and gain insights. For Zume, that means making the food supply chain more efficient. When we talk about last mile delivery, we’re actually talking about the last five or seven mile delivery. Zume is using predictive analytics to know ahead of time what food will be ordered, when and where. From that information they can place mobile kitchens directly in those neighborhoods to make the delivery process more efficient for the cooks, the couriers and the consumer.

Satchell wants to take this type of AI-based prediction up the food stack to improve supply chains and even create predictive farming. This type of just-in-time farming would help farmers understand what types of crops to grow and when in order to help reduce food waste.

As Andrejko pointed out in his talk, agriculture is already changing thanks to analytics and AI. Though Andrejko would like people to think of AI as “Augmented Intelligence,” not “Artificial Intelligence.” On the farm, this means that data and algorithms can be used to optimize how fertilizer is applied, using more on acres that need it and less where it doesn’t. Or with emerging fruit-picking robots that can use computer vision to automatically harvest at peak times for ripeness.

Andrejko also sees augmented intelligence at our kitchen tables, giving us more human connection at dinner time. Say you want to cook a butternut squash chili for dinner. Eventually you’ll be able to say that request to a voice assistant, which will break down the list of ingredients and place the order, which will be brought to your house via a self-driving delivery vehicle, which also uses AI to travel to your door.

These talks, along with the panel discussion with Satchell and Andrejko afterwards, are great deep dives into AI for anyone curious about the trendy term or for any company looking to add that arrow to their quiver. At least have the marketing department in your life check it out.

The post SKS 2019: Think of AI as Augmented Intelligence, and is the Future Just-in-Time Farming? appeared first on Artificial Intelligence.

• Executes projects in the field of advanced automated horticultural production systems with a focus on computer vision and machine learning applications; contributes to the development of sustainable (indoor) horticultural crop production systems.
• Develops new image analysis and control software for the detection of crop growth parameters of various crops, the quality detection of fresh products and crops and the detection of plant health and resilience; uses various camera techniques such as RGB, 3D, hyperspectral or thermal imaging.
• Applies modern machine learning methods such as deep learning and reinforcement learning, develops new AI algorithms for growing horticultural crops with a minimum of resources; deals with time series; uses various sensors; deals with data fusion.
• Leads and works in projects together with other academic and industrial partners.
• Conducts applied scientific research within the projects.
• Communicates relevant research results to science and practise.

Requirements:

We are looking for an enthusiastic and ambitious researcher computer vision and machine learning, with:

• Academic working and thinking on the field of computer vision and machine learning, study in e.g. artificial intelligence, natural or biomedical science, holds a PhD.
• Knowledge of various image analysis techniques, data analysis methods, classification and statistics.
• Knowledge of various modern machine learning techniques, such as deep learning and reinforcement learning methods.
• Experience with the structured development of complex software, programming in C++ and C#; experience with Halcon and / or LabView and / or Matlab is an advantage.
• Knowledge of lighting, optics and camera systems (2D, 3D) including calibration and control is an advantage; knowledge of various sensor types, data output and practical usage of such sensors is an advantage.
• Experience with the application of image processing and machine learning in the green environment and willingness to learn the challenges of horticulture.
• Willingness to combine theoretical knowledge with practical setups and implementation.
• Experience with project-based work and project leadership.
• Excellent knowledge of Dutch and English language.
• Driving License B.
• Is an enthusiastic independent researcher with good communication skills and team spirit.
• Is customer- and quality-oriented and stress tolerant.

Salary Benefits:

A challenging position with, depending on your experience, a competitive salary up to a maximum of € 4.934,- gross per month for a full working week of 36 hours in accordance with the Collective Labor Agreement for Wageningen Research (scale 11). At WUR you work in a (inter-)national leading organization in the field of research and education. This is a position for a duration of one year with the prospect of permanent employment if our cooperation is mutually satisfactory.

In addition, we offer:

• 8% holiday allowance;
• a fixed end-of-year bonus of 3%;
• excellent training opportunities and secondary employment conditions;
• flexible working hours and vacations can be determined in consultation with colleagues in such a way that an optimal balance between work and private life is possible;
• excellent pension plan through ABP;
• 171 vacation hours, the possibility to purchase extra and good supplementary leave schemes;
• a choice model to put together part of your employment conditions yourself, such as a bicycle plan;
• make use of the sports facilities on campus for a small fee.

Wageningen University & Research stimulates internal career opportunities and mobility with an active internal recruitment policy. There are ample opportunities for own initiative in a learning environment. We offer a versatile job in an international environment with varied activities in a pleasant and open working atmosphere.

The post Researcher Computer Vision en Machine Learning appeared first on Artificial Intelligence.

Will future computers be conscious entities? Will they have free will? Or will they just be simulating these capabilities?

A popular argument against computers being able to think in a way analogous to humans goes like this.

We humans are conscious beings and have free will and these are essential to our thinking. Computers are mechanisms which run the same way every time and therefore cannot have free will. Computers are made of materials which cannot possess the essence of consciousness. Therefore, without free will and consciousness, computers will never be able to think.

With the discussion of free will and consciousness, we have reached the pinnacle of human mental processes and also a point of philosophical discussion.

To me, the question devolves into one of whether or not you accept modern physics as describing reality. While there are certainly areas of physics which are yet to be discovered, the essential point is that any physical system can be represented by information which can be replicated in computers. If what we observe in the electrochemistry of the brain includes consciousness and free will, then there is no reason a computer cannot equally possess these capabilities as well. If we believe there is some unobserved essential “magic”, then we have a choice. Either the magic will eventually be observed, defined as a part of physics and included in computers OR the magic is beyond the scope of observation, meaning it is outside the scope of any future conceivable physics.

My contention is that human thought is the sum total of a multiplicity of general mental functions working in parallel on an unimaginably large scale. These functions were presented in the last few chapters, and each function can be described and understood. Some of these functions are already working in computers at levels higher than in humans and those that are not, conceivably, could be in the near future.

I contend that all the functions presented so far are necessary to AGI and are also necessary to any appearance of consciousness. Without the “sensation” of the world and the modeling and imagination necessary to comprehend it, no AI system could ever put its chess game, mathematical proof or car-driving skill in context—the context being a real-world environment.

I further contend that the functions presented so far are sufficient for AGI to have the appearance of consciousness. With these capabilities, a robot or computer with appropriate peripherals could sense its surroundings, remember previous situations, and learn how various situations affect it. It could simulate several possible actions at any given time and select and perform the one it determined was best.

We already have computers which can speak and understand speech to some extent. Coupled with a robotic “body” with vision, such a system could learn about objects in its environment and appear to reason. It will appear to make reasoned decisions and will be able to explain its rationale.

Free will

Given its simulation capabilities, your mind is able to select several choices, play them out somewhat through simulation, then select the one which results in your most beneficial outcome.

If you were to try to prove to yourself that you have free will, you might place yourself repeatedly in a situation which is as identical as possible. Then sometimes you make one choice and subsequently a different choice. Unfortunately, you can never place yourself in a truly identical situation because after the first time, your experience, your choice, and the outcome all become a part of your mind and so the state of your mind is different on the next try. As a result, we do not have any method of measuring whether or not free will actually exists because we can never set up truly identical situations to determine if we could make different choices.

Here’s a demonstration. Raise your right index finger.

Did you raise it? If you did, was it simply to play along with my demonstration in hopes of learning something? Or if you didn’t raise it, was it because you wanted to assert your “free will?” I contend that whatever decision you made, it depended completely on your current state of mind — based on your experience with similar demonstrations.

OK, now raise your left index finger. Did you do the opposite of what you did on the previous paragraph? Did you assert your free will? Either way, did you think about raising your finger? I bet you did. I bet that when you read the text, you couldn’t avoid thinking about it.

There is no way to prove or disprove free will in either instance. Your first decision is based on your previous experience and your second decision is based on the same experience plus your experience with the first decision.

When computers become learning systems they will likewise incorporate the experience of a past decision into the process of making a present decision. So the future computer, left to its normal operation, may make a different choice when re-encountering an identical situation, just as you might.

On the other hand, with computers we can set up situations which are truly identical. Computers can be restarted to the specific point of their previous backup so their previous experience need not become a part of their present operation. Restoring a backup can completely erase the experience of the first decision.

So if you were to make a backup of the computer’s entire state, have the computer make a decision, reload it from the backup again, put it in the identical situation and let it make the decision again, it would always make the same decision. If it did not, we would consider this a malfunction. One of the convenient things about computer situations is that we can control all the inputs (including access to real-time clocks) to make the situation absolutely identical.

There are theories of human free will and consciousness which rely on complex mechanisms or quantum mechanics and these may eventually be shown to be relevant. The simpler theory, as uncomfortable as it seems, is that the human’s free will is just like the learning computer’s. It is simply that we can never set up identical situations for ourselves and so we cannot test if the theory is correct. We each only make the choice for the best expected outcome for each situation we encounter.

We see from examining the brain and the operation of neurons that there is nothing observable in the brain which makes it appear to be detached from the laws of physics. The laws of physics are deterministic until you reach the level of (very small) quantum particles. You might argue that the human brain would make a different choice in a truly identical situation because its computing mechanism is governed by unpredictable quantum mechanics and chaos theory. That our synapses may send a few more, or a few less, molecules of neurotransmitters, depending on quantum effects. Therefore we might reach a different conclusion for an identical situation. For me, this is an unsatisfying argument because it only replaces the concept of deterministic “free will” with the free will of a roulette wheel. It is disquieting enough to believe that your mind is a deterministic mechanism without saying it’s unreliable as well (and then going further by claiming the mind’s superiority because of its unreliability).

Google search always returns its best search results (ignoring sponsorship). You might disagree with the algorithmic definition of “best” but Google computers can only do what they are directed to do. However, if users never click on the top search entry, it will eventually be de-rated and appear further down the list. So, for a given search request, you could theoretically get a different search result every time the search was requested as Google’s servers attempt to please you—producing the “best” result. Google computers are incorporating the qualitative experience of a specific search result into ranking decisions for future searches—whether or not it pleased its users—by putting a certain result at the top.

Are Google’s search servers aware that they have free will? Of course not. Do they actually have free will in the same context that humans do?

Think of it this way. Are your decisions different because you believe you have free will? Absolutely. One of your mind’s innate objectives seems to be to assert its own individuality. Google’s search computers don’t consider the possibility of presenting different results in order simply to demonstrate their free will (as you might have with your index fingers above). It seems obvious that one of the inputs to any decision you make is your belief in your ability to make a decision… your belief that you have free will.

Because your belief in free will is another input into the decision process, you would likely make different decisions if you didn’t believe you had free will. If you really don’t believe you have free will, why would you ponder making any decision at all? You’d be a purely reactive entity.

The reason to ponder a decision (the reason to believe in free will) is the probability of making a better decision by examining the ramifications of different possibilities. Your brain doesn’t seem to have the ability to simulate different possibilities simultaneously, so it examines them one at a time. The process of examining different possibilities leads you to the belief in free will and the belief leads you to examine more possibilities.

But belief requires consciousness…

The post Hitting the books: Will Computers Revolt? appeared first on Artificial Intelligence.