Daily Archives: August 5, 2017

Over the past couple of weeks, youve likely seen one of your Facebook friends sharing an article about how a pair of AI-driven Facebook chatbots invented their own language in a deviation from what they were originally programmed to do. The gist is this: Facebook created twin AI chat programs to converse with each other (and learn from each other), and they eventually stopped communicating in English and began communicating in a non-English language they invented.

The headlines reporting about this range from dramatized to exploitative, such as the Telegraphs Facebook shuts down robots after they invent their own language, making it sound like these chatbots were conspiring against humanity or posing some other existential threat. A quick look at any comment feed will show you users responding with fear, excitement, and amusement, with lines like And so it begins or references to the works of Isaac Asimov.

But as Facebooks AI Dhruv Batra noted on Monday, AIs have been inventing new ways of communicating with each other for decades, so the news was, in fact, not news. Batra also explained that, contrary to the headlines, the experiment wasnt shut down but simply altered to tweak the linguistic exchange.

Furthermore, there has been no attempt to hide details of the experiment. Everything is out in the open, with all details publicly available on GitHub, allowing other coders replicate the scenario.

The bottom line is this: New robot languages should be the least of our concerns when it comes to AI.

Tech-minded influencers like Elon Musk and Bill Gates have donated significant time and money to explain AI is a threat and figure out new ways to advance it ethically.

These are some of the main areas where we should be focusing our attention:

Weapons. Military drones are already in operation, and autonomous weaponry has been described as the third revolution in warfare. Automated intelligent weaponry puts fewer soldiers in harms way and is far less expensive than other advanced technologies like fighter jets or nuclear materials. For example, current Reaper drones cost about $13 million, compared to $100 million for a fighter jet. In the future, theyll become far cheaper and ubiquitously available in unregulated (and that means incredibly powerful) mobile weapons that anyone can program to kill and destroy (unless we proactively impose security measures).

Control. What may be most important isnt the AI itself, but whos controlling it. Were dealing with forces that have the power to reshape our world, and if theyre monopolized by greedy corporations, power-hungry nations, or even well-intentioned individuals who simply dont know what theyre doing, the technology could easily be abused or misused. Thats why several Silicon Valley influencers and AI researchers have come together to form OpenAI, an initiative to make AI available to the entire world, not just one group of people. The initiative hasnt been cheap, with one top AI researcher reportedly being offered two to three times his market value (which is already more than a starting NFL quarterback).

Growth. The other problem with AI is the pace of its growth, when uncontrolled by limiting parameters. Artificial general intelligence (AGI) has the power to improve exponentially, since it will hypothetically be able to improve itself, and at some point it will surpass the intelligence of its creators. The agricultural age lasted millennia, the industrial age lasted centuries, the information age has lasted decades, and now the age of AI may last mere years; technology growth accelerates rather than progressing linearly, and now were at a point where any further acceleration will leave us woefully unprepared to deal with the consequences (if we take a reactive, rather than a proactive stance). We havent been able to create superintelligence yet, but we already have the processing power in place the Tianhe-2 in China is capable of 34 quadrillion computations per second (cps), far more than the human brain (at 10 quadrillion cps). If we want to wield this computational power responsibly, we need to have a foundation in place to deal with its unrestricted growth. That means understanding the ethics of AI development, how limitations could work, and having safeguards in place in case something goes awry.

There are some legitimate and serious existential concerns about how AI is being used and how it may develop in the future. However, its irresponsible to overly personify the systems being tested or allow headlines to shape our perspectives. Chatbots inventing a new language isnt threatening; its natural, and our attention belongs elsewhere.

Tony Tie is senior search marketer at Expedia. He has previously worked with a number of Fortune 500 companies to improve their online presence. He is also a marketing and entrepreneurship lecturer at various universities.

Link:

New AI languages should be the least of our concerns - VentureBeat

When 16-year-old Kavya Kopparapu wasnt attending conferences, giving speeches, presiding over her schools bioinformatics society, organizing a research symposium, playing piano, and running a non-profit, she worried about what to do with all her free time.

It was June 2016, the summer after her junior year in high school, and Kopparapu was looking for a new project that would use her computer science skills. Her thoughts quickly turned to her grandfather, who lives in a small city on Indias eastern coast.

In 2013 he began showing symptoms of diabetic retinopathy, a complication of diabetes that damages blood vessels in the retina and can lead to blindness. Eventually he was diagnosed and treated, but not before his vision deteriorated. Still, he was lucky: Although treatments such as medication and surgery can stop or even reverse eye damage if the disease is caught early, most patients never receive care.

Kopparapu knows the statistics by heart: Of 415 million diabetics worldwide, one-third will develop retinopathy. Fifty percent will be undiagnosed. Of patients with severe forms, half will go blind in five years. Most will be poor.

The lack of diagnosis is the biggest challenge, Kopparapu says. In India, there are programs that send doctors into villages and slums, but there are a lot of patients and only so many ophthalmologists. What if there were a cheap, easy way for local clinicians to find new cases and refer them to a hospital?

That was the genesis of Eyeagnosis, a smartphone app plus 3D-printed lens that seeks to change the diagnostic procedure from a 2-hour exam requiring a multi-thousand-dollar retinal imager to a quick photo snap with a phone.

Kopparapu and her teamincluding her 15-year-old brother, Neeyanth, and her high school classmate Justin Zhangtrained an artificial intelligence system to recognize signs of diabetic retinopathy in photos of eyes and offer a preliminary diagnosis. She presented the system at the OReilly Artificial Intelligence conference, in New York City, last month.

The device is ideal for making screening much more efficient and available to a broader population, says J. Fielding Hejtmancik, an expert in visual diseases at the National Institutes of Health (NIH). Other research groups, including Googleand Peek Vision, have recently announced similar systems, but Hejtmancik is impressed with the students ingenuity. These kids have put things together in a very nice way thats a bit cheaper and simpler than most [systems designed by researchers]who, by the way, all have advanced degrees!

Kopparapu has always had a scientific mind. Growing up in Herndon, Virginia, she and her brotherbuilt Knex creations, watched MythBusters and Cosmos, and read Scientific American together over breakfast. But she didnt get hooked on computers until she attended a programming workshop run by the National Center for Women and Information Technology. I went home and taught myself Java, HTML, Python, C, she says. My mom had to tear me away from the computer. Id forget to eat.

In high school, she took classes on computer science, then computer vision, then artificial intelligencebut she was troubled to realized that in each class, she was one of only a few girls. She resolved to start an organization to empower girls to pursue computer science. I dont think the problem is a lack of passion, she says. Its more I dont feel like Im good enough. She founded the Girls Computing League, wooed sponsors such as Amazon Web Services and the president of Harvey Mudd College, and now puts on coding workshops for underprivileged kids.

Eyeagnosis began as most endeavors do these days. I googled a lot, Kopparapu says. She also sent a lot of emailsto ophthalmologists, computational pathologists, biochemists, epidemiologists, neuroscientists, physicists, and experts in machine learning. Then she put together a plan.

First her team worked on the diagnostic AI, choosing to use a machine-learning architecture known as a convolutional neural network (CNN). Neural nets are behind the recent explosion in artificial intelligence, including advances in speech recognition, machine translation, and image captioning. They acquire these skills by parsing vast sets of data (millions of photos of cats, say) and looking for patterns of similarity.

CNNs are especially good at classifying images, so its no coincidence that their design resembles the brains visual system. Information passes through hierarchical layers of neurons called nodes; with each layer, the network recognizes ever more abstract features: Pixels become edges become shapes become objects. Its kind of funny that were using a system based on how the retinal system works to diagnose a retinal disease, Kopparapu says.

Rather than build a network from scratch, she chose an off-the-shelf model developed by Microsoft researchers called ResNet-50. But in order to teach the system to recognize an eye disease, she needed training data.

She found that data in the NIHs EyeGene database, which included 34,000 retinal scans. Many of these images, taken under various conditions with different types of cameras, were blurry or poorly exposed. But that was actually a good thing, Kopporapu says. Its very representative of the real-world conditions youd get with using a smartphone.

By August 2016, her team had trained ResNet-50 to spot diabetic retinopathy with the accuracy of a human pathologist. By October, she had made arrangements with Aditya Jyot Eye Hospital, in Mumbai, to test the Eyeagnosis app, which not only detects disease but also highlights blood vessels and microaneurysms in an imagea process that normally involves injecting a fluorescent dye into patients blood. Were trying to make it as easy as possible for an ophthalmologist to look at all that info and say Heres my final diagnosis.

In November, she shipped her first 3D-printed prototype for the systems lens to the hospital. When fitted onto a smartphone, the lens focuses the phones diffuse, off-centered flash to best illuminate a retina. The complete Eyeagnosis system has already been tried on five patients at the hospital, and in each case it made an accurate diagnosis.

Hejtmancik, the NIH expert, notes that theres a long road to clinical adoption. What shes going to need is a lot of clinical data showing that [Eyeagnosis] is reliable under a variety of situations: in eye hospitals, in the countryside, in clinics out in the boonies of India, he says.

Still, Hejtmancik thinks the system has real commercial potential. The only problem, he says, is that its so cheap that big companies might not see the potential for a profit margin. But that affordability is exactly what you want in medical care, in my opinion, he says.

IEEE Spectrums biomedical engineering blog, featuring the wearable sensors, big data analytics, and implanted devices that enable new ventures in personalized medicine.

Sign up for The Human OS newsletter and get biweekly news about how technology is making healthcare smarter.

An artificial intelligence program correctly identifies 355 more patients who developed cardiovascular disease 1May

Algorithms spot heart failure, lung cancer, and drug candidates sooner 5Apr

Ray Kurzweil, Rodney Brooks, and others weigh in on the future of artificial intelligence 31May

Smartphone-based diagnostic tools with an artificial intelligence upgrade could save millions of lives 27Apr

The ear-worn sensor's current version continuously measures core body temperature and acts as a hearing aid 27Jul

Researchers build the most complex RNA-based computer in living bacterial cells 26Jul

Technique for organizing DNA molecules closer together on chip boosts processing speeds 24Jul

Gold nanomesh sensors last for a week on the skin without causing inflammation 20Jul

A machine-learning algorithm picked up differences in brain connections in kids as young as six months old 7Jun

John Rogers's stick-on sensors could keep neonates comfortable 22May

Ford is experimenting with car features that could help drivers with diabetes, heart problems, and more 19May

EEG readings of brain activity could help study pain relief treatments for babies who cannot speak 3May

MITs WiGait wall sensor can unobtrusively monitor people for many health conditions based on their walking patterns 1May

Engineered cells implanted in diabetic mice produce insulin on demand 26Apr

Final Frontier Medical Devices takes home $2.6 million for mobile diagnostic device 14Apr

This lab-on-a-glove puts weapons detection on the user's fingertips 30Mar

Cows and people are the large animals that most dominate the environment 29Mar

Harvard's home testing kit for male fertility would just need your smartphone and semen sample 23Mar

These artificial intelligence systems could smarten up intensive care units 7Mar

This sleek yet rugged sensor measures better and lasts longer 27Feb

Go here to read the rest:

Teenage Whiz Kid Invents an AI System to Diagnose Her Grandfather's Eye Disease - IEEE Spectrum