Tag Archives: see
#431238 AI Is Easy to Fool—Why That Needs to ...
Con artistry is one of the world’s oldest and most innovative professions, and it may soon have a new target. Research suggests artificial intelligence may be uniquely susceptible to tricksters, and as its influence in the modern world grows, attacks against it are likely to become more common.
The root of the problem lies in the fact that artificial intelligence algorithms learn about the world in very different ways than people do, and so slight tweaks to the data fed into these algorithms can throw them off completely while remaining imperceptible to humans.
Much of the research into this area has been conducted on image recognition systems, in particular those relying on deep learning neural networks. These systems are trained by showing them thousands of examples of images of a particular object until they can extract common features that allow them to accurately spot the object in new images.
But the features they extract are not necessarily the same high-level features a human would be looking for, like the word STOP on a sign or a tail on a dog. These systems analyze images at the individual pixel level to detect patterns shared between examples. These patterns can be obscure combinations of pixel values, in small pockets or spread across the image, that would be impossible to discern for a human, but highly accurate at predicting a particular object.
“An attacker can trick the object recognition algorithm into seeing something that isn’t there, without these alterations being obvious to a human.”
What this means is that by identifying these patterns and overlaying them over a different image, an attacker can trick the object recognition algorithm into seeing something that isn’t there, without these alterations being obvious to a human. This kind of manipulation is known as an “adversarial attack.”
Early attempts to trick image recognition systems this way required access to the algorithm’s inner workings to decipher these patterns. But in 2016 researchers demonstrated a “black box” attack that enabled them to trick such a system without knowing its inner workings.
By feeding the system doctored images and seeing how it classified them, they were able to work out what it was focusing on and therefore generate images they knew would fool it. Importantly, the doctored images were not obviously different to human eyes.
These approaches were tested by feeding doctored image data directly into the algorithm, but more recently, similar approaches have been applied in the real world. Last year it was shown that printouts of doctored images that were then photographed on a smartphone successfully tricked an image classification system.
Another group showed that wearing specially designed, psychedelically-colored spectacles could trick a facial recognition system into thinking people were celebrities. In August scientists showed that adding stickers to stop signs in particular configurations could cause a neural net designed to spot them to misclassify the signs.
These last two examples highlight some of the potential nefarious applications for this technology. Getting a self-driving car to miss a stop sign could cause an accident, either for insurance fraud or to do someone harm. If facial recognition becomes increasingly popular for biometric security applications, being able to pose as someone else could be very useful to a con artist.
Unsurprisingly, there are already efforts to counteract the threat of adversarial attacks. In particular, it has been shown that deep neural networks can be trained to detect adversarial images. One study from the Bosch Center for AI demonstrated such a detector, an adversarial attack that fools the detector, and a training regime for the detector that nullifies the attack, hinting at the kind of arms race we are likely to see in the future.
While image recognition systems provide an easy-to-visualize demonstration, they’re not the only machine learning systems at risk. The techniques used to perturb pixel data can be applied to other kinds of data too.
“Bypassing cybersecurity defenses is one of the more worrying and probable near-term applications for this approach.”
Chinese researchers showed that adding specific words to a sentence or misspelling a word can completely throw off machine learning systems designed to analyze what a passage of text is about. Another group demonstrated that garbled sounds played over speakers could make a smartphone running the Google Now voice command system visit a particular web address, which could be used to download malware.
This last example points toward one of the more worrying and probable near-term applications for this approach: bypassing cybersecurity defenses. The industry is increasingly using machine learning and data analytics to identify malware and detect intrusions, but these systems are also highly susceptible to trickery.
At this summer’s DEF CON hacking convention, a security firm demonstrated they could bypass anti-malware AI using a similar approach to the earlier black box attack on the image classifier, but super-powered with an AI of their own.
Their system fed malicious code to the antivirus software and then noted the score it was given. It then used genetic algorithms to iteratively tweak the code until it was able to bypass the defenses while maintaining its function.
All the approaches noted so far are focused on tricking pre-trained machine learning systems, but another approach of major concern to the cybersecurity industry is that of “data poisoning.” This is the idea that introducing false data into a machine learning system’s training set will cause it to start misclassifying things.
This could be particularly challenging for things like anti-malware systems that are constantly being updated to take into account new viruses. A related approach bombards systems with data designed to generate false positives so the defenders recalibrate their systems in a way that then allows the attackers to sneak in.
How likely it is that these approaches will be used in the wild will depend on the potential reward and the sophistication of the attackers. Most of the techniques described above require high levels of domain expertise, but it’s becoming ever easier to access training materials and tools for machine learning.
Simpler versions of machine learning have been at the heart of email spam filters for years, and spammers have developed a host of innovative workarounds to circumvent them. As machine learning and AI increasingly embed themselves in our lives, the rewards for learning how to trick them will likely outweigh the costs.
Image Credit: Nejron Photo / Shutterstock.com Continue reading
Posted in Human Robots
#431186 The Coming Creativity Explosion Belongs ...
Does creativity make human intelligence special?
It may appear so at first glance. Though machines can calculate, analyze, and even perceive, creativity may seem far out of reach. Perhaps this is because we find it mysterious, even in ourselves. How can the output of a machine be anything more than that which is determined by its programmers?
Increasingly, however, artificial intelligence is moving into creativity’s hallowed domain, from art to industry. And though much is already possible, the future is sure to bring ever more creative machines.
What Is Machine Creativity?
Robotic art is just one example of machine creativity, a rapidly growing sub-field that sits somewhere between the study of artificial intelligence and human psychology.
The winning paintings from the 2017 Robot Art Competition are strikingly reminiscent of those showcased each spring at university exhibitions for graduating art students. Like the works produced by skilled artists, the compositions dreamed up by the competition’s robotic painters are aesthetically ambitious. One robot-made painting features a man’s bearded face gazing intently out from the canvas, his eyes locking with the viewer’s. Another abstract painting, “inspired” by data from EEG signals, visually depicts the human emotion of misery with jagged, gloomy stripes of black and purple.
More broadly, a creative machine is software (sometimes encased in a robotic body) that synthesizes inputs to generate new and valuable ideas, solutions to complex scientific problems, or original works of art. In a process similar to that followed by a human artist or scientist, a creative machine begins its work by framing a problem. Next, its software specifies the requirements the solution should have before generating “answers” in the form of original designs, patterns, or some other form of output.
Although the notion of machine creativity sounds a bit like science fiction, the basic concept is one that has been slowly developing for decades.
Nearly 50 years ago while a high school student, inventor and futurist Ray Kurzweil created software that could analyze the patterns in musical compositions and then compose new melodies in a similar style. Aaron, one of the world’s most famous painting robots, has been hard at work since the 1970s.
Industrial designers have used an automated, algorithm-driven process for decades to design computer chips (or machine parts) whose layout (or form) is optimized for a particular function or environment. Emily Howell, a computer program created by David Cope, writes original works in the style of classical composers, some of which have been performed by human orchestras to live audiences.
What’s different about today’s new and emerging generation of robotic artists, scientists, composers, authors, and product designers is their ubiquity and power.
“The recent explosion of artificial creativity has been enabled by the rapid maturation of the same exponential technologies that have already re-drawn our daily lives.”
I’ve already mentioned the rapidly advancing fields of robotic art and music. In the realm of scientific research, so-called “robotic scientists” such as Eureqa and Adam and Eve develop new scientific hypotheses; their “insights” have contributed to breakthroughs that are cited by hundreds of academic research papers. In the medical industry, creative machines are hard at work creating chemical compounds for new pharmaceuticals. After it read over seven million words of 20th century English poetry, a neural network developed by researcher Jack Hopkins learned to write passable poetry in a number of different styles and meters.
The recent explosion of artificial creativity has been enabled by the rapid maturation of the same exponential technologies that have already re-drawn our daily lives, including faster processors, ubiquitous sensors and wireless networks, and better algorithms.
As they continue to improve, creative machines—like humans—will perform a broad range of creative activities, ranging from everyday problem solving (sometimes known as “Little C” creativity) to producing once-in-a-century masterpieces (“Big C” creativity). A creative machine’s outputs could range from a design for a cast for a marble sculpture to a schematic blueprint for a clever new gadget for opening bottles of wine.
In the coming decades, by automating the process of solving complex problems, creative machines will again transform our world. Creative machines will serve as a versatile source of on-demand talent.
In the battle to recruit a workforce that can solve complex problems, creative machines will put small businesses on equal footing with large corporations. Art and music lovers will enjoy fresh creative works that re-interpret the style of ancient disciplines. People with a health condition will benefit from individualized medical treatments, and low-income people will receive top-notch legal advice, to name but a few potentially beneficial applications.
How Can We Make Creative Machines, Unless We Understand Our Own Creativity?
One of the most intriguing—yet unsettling—aspects of watching robotic arms skillfully oil paint is that we humans still do not understand our own creative process. Over the centuries, several different civilizations have devised a variety of models to explain creativity.
The ancient Greeks believed that poets drew inspiration from a transcendent realm parallel to the material world where ideas could take root and flourish. In the Middle Ages, philosophers and poets attributed our peculiarly human ability to “make something of nothing” to an external source, namely divine inspiration. Modern academic study of human creativity has generated vast reams of scholarship, but despite the value of these insights, the human imagination remains a great mystery, second only to that of consciousness.
Today, the rise of machine creativity demonstrates (once again), that we do not have to fully understand a biological process in order to emulate it with advanced technology.
Past experience has shown that jet planes can fly higher and faster than birds by using the forward thrust of an engine rather than wings. Submarines propel themselves forward underwater without fins or a tail. Deep learning neural networks identify objects in randomly-selected photographs with super-human accuracy. Similarly, using a fairly straightforward software architecture, creative software (sometimes paired with a robotic body) can paint, write, hypothesize, or design with impressive originality, skill, and boldness.
At the heart of machine creativity is simple iteration. No matter what sort of output they produce, creative machines fall into one of three categories depending on their internal architecture.
Briefly, the first group consists of software programs that use traditional rule-based, or symbolic AI, the second group uses evolutionary algorithms, and the third group uses a variation of a form of machine learning called deep learning that has already revolutionized voice and facial recognition software.
1) Symbolic creative machines are the oldest artificial artists and musicians. In this approach—also known as “good old-fashioned AI (GOFAI) or symbolic AI—the human programmer plays a key role by writing a set of step-by-step instructions to guide the computer through a task. Despite the fact that symbolic AI is limited in its ability to adapt to environmental changes, it’s still possible for a robotic artist programmed this way to create an impressively wide variety of different outputs.
2) Evolutionary algorithms (EA) have been in use for several decades and remain powerful tools for design. In this approach, potential solutions “compete” in a software simulator in a Darwinian process reminiscent of biological evolution. The human programmer specifies a “fitness criterion” that will be used to score and rank the solutions generated by the software. The software then generates a “first generation” population of random solutions (which typically are pretty poor in quality), scores this first generation of solutions, and selects the top 50% (those random solutions deemed to be the best “fit”). The software then takes another pass and recombines the “winning” solutions to create the next generation and repeats this process for thousands (and sometimes millions) of generations.
3) Generative deep learning (DL) neural networks represent the newest software architecture of the three, since DL is data-dependent and resource-intensive. First, a human programmer “trains” a DL neural network to recognize a particular feature in a dataset, for example, an image of a dog in a stream of digital images. Next, the standard “feed forward” process is reversed and the DL neural network begins to generate the feature, for example, eventually producing new and sometimes original images of (or poetry about) dogs. Generative DL networks have tremendous and unexplored creative potential and are able to produce a broad range of original outputs, from paintings to music to poetry.
The Coming Explosion of Machine Creativity
In the near future as Moore’s Law continues its work, we will see sophisticated combinations of these three basic architectures. Since the 1950s, artificial intelligence has steadily mastered one human ability after another, and in the process of doing so, has reduced the cost of calculation, analysis, and most recently, perception. When creative software becomes as inexpensive and ubiquitous as analytical software is today, humans will no longer be the only intelligent beings capable of creative work.
This is why I have to bite my tongue when I hear the well-intended (but shortsighted) advice frequently dispensed to young people that they should pursue work that demands creativity to help them “AI-proof” their futures.
Instead, students should gain skills to harness the power of creative machines.
There are two skills in which humans excel that will enable us to remain useful in a world of ever-advancing artificial intelligence. One, the ability to frame and define a complex problem so that it can be handed off to a creative machine to solve. And two, the ability to communicate the value of both the framework and the proposed solution to the other humans involved.
What will happen to people when creative machines begin to capably tread on intellectual ground that was once considered the sole domain of the human mind, and before that, the product of divine inspiration? While machines engaging in Big C creativity—e.g., oil painting and composing new symphonies—tend to garner controversy and make the headlines, I suspect the real world-changing application of machine creativity will be in the realm of everyday problem solving, or Little C. The mainstream emergence of powerful problem-solving tools will help people create abundance where there was once scarcity.
Image Credit: adike / Shutterstock.com Continue reading
Posted in Human Robots
#431159 How Close Is Turing’s Dream of ...
The quest for conversational artificial intelligence has been a long one.
When Alan Turing, the father of modern computing, racked his considerable brains for a test that would truly indicate that a computer program was intelligent, he landed on this area. If a computer could convince a panel of human judges that they were talking to a human—if it could hold a convincing conversation—then it would indicate that artificial intelligence had advanced to the point where it was indistinguishable from human intelligence.
This gauntlet was thrown down in 1950 and, so far, no computer program has managed to pass the Turing test.
There have been some very notable failures, however: Joseph Weizenbaum, as early as 1966—when computers were still programmed with large punch-cards—developed a piece of natural language processing software called ELIZA. ELIZA was a machine intended to respond to human conversation by pretending to be a psychotherapist; you can still talk to her today.
Talking to ELIZA is a little strange. She’ll often rephrase things you’ve said back at you: so, for example, if you say “I’m feeling depressed,” she might say “Did you come to me because you are feeling depressed?” When she’s unsure about what you’ve said, ELIZA will usually respond with “I see,” or perhaps “Tell me more.”
For the first few lines of dialogue, especially if you treat her as your therapist, ELIZA can be convincingly human. This was something Weizenbaum noticed and was slightly alarmed by: people were willing to treat the algorithm as more human than it really was. Before long, even though some of the test subjects knew ELIZA was just a machine, they were opening up with some of their deepest feelings and secrets. They were pouring out their hearts to a machine. When Weizenbaum’s secretary spoke to ELIZA, even though she knew it was a fairly simple computer program, she still insisted Weizenbaum leave the room.
Part of the unexpected reaction ELIZA generated may be because people are more willing to open up to a machine, feeling they won’t be judged, even if the machine is ultimately powerless to do or say anything to really help. The ELIZA effect was named for this computer program: the tendency of humans to anthropomorphize machines, or think of them as human.
Weizenbaum himself, who later became deeply suspicious of the influence of computers and artificial intelligence in human life, was astonished that people were so willing to believe his script was human. He wrote, “I had not realized…that extremely short exposures to a relatively simple computer program could induce powerful delusional thinking in quite normal people.”
“Consciously, you know you’re talking to a big block of code stored somewhere out there in the ether. But subconsciously, you might feel like you’re interacting with a human.”
The ELIZA effect may have disturbed Weizenbaum, but it has intrigued and fascinated others for decades. Perhaps you’ve noticed it in yourself, when talking to an AI like Siri, Alexa, or Google Assistant—the occasional response can seem almost too real. Consciously, you know you’re talking to a big block of code stored somewhere out there in the ether. But subconsciously, you might feel like you’re interacting with a human.
Yet the ELIZA effect, as enticing as it is, has proved a source of frustration for people who are trying to create conversational machines. Natural language processing has proceeded in leaps and bounds since the 1960s. Now you can find friendly chatbots like Mitsuku—which has frequently won the Loebner Prize, awarded to the machines that come closest to passing the Turing test—that aim to have a response to everything you might say.
In the commercial sphere, Facebook has opened up its Messenger program and provided software for people and companies to design their own chatbots. The idea is simple: why have an app for, say, ordering pizza when you can just chatter to a robot through your favorite messenger app and make the order in natural language, as if you were telling your friend to get it for you?
Startups like Semantic Machines hope their AI assistant will be able to interact with you just like a secretary or PA would, but with an unparalleled ability to retrieve information from the internet. They may soon be there.
But people who engineer chatbots—both in the social and commercial realm—encounter a common problem: the users, perhaps subconsciously, assume the chatbots are human and become disappointed when they’re not able to have a normal conversation. Frustration with miscommunication can often stem from raised initial expectations.
So far, no machine has really been able to crack the problem of context retention—understanding what’s been said before, referring back to it, and crafting responses based on the point the conversation has reached. Even Mitsuku will often struggle to remember the topic of conversation beyond a few lines of dialogue.
“For everything you say, there could be hundreds of responses that would make sense. When you travel a layer deeper into the conversation, those factors multiply until you end up with vast numbers of potential conversations.”
This is, of course, understandable. Conversation can be almost unimaginably complex. For everything you say, there could be hundreds of responses that would make sense. When you travel a layer deeper into the conversation, those factors multiply until—like possible games of Go or chess—you end up with vast numbers of potential conversations.
But that hasn’t deterred people from trying, most recently, tech giant Amazon, in an effort to make their AI voice assistant, Alexa, friendlier. They have been running the Alexa Prize competition, which offers a cool $500,000 to the winning AI—and a bonus of a million dollars to any team that can create a ‘socialbot’ capable of sustaining a conversation with human users for 20 minutes on a variety of themes.
Topics Alexa likes to chat about include science and technology, politics, sports, and celebrity gossip. The finalists were recently announced: chatbots from universities in Prague, Edinburgh, and Seattle. Finalists were chosen according to the ratings from Alexa users, who could trigger the socialbots into conversation by saying “Hey Alexa, let’s chat,” although the reviews for the socialbots weren’t always complimentary.
By narrowing down the fields of conversation to a specific range of topics, the Alexa Prize has cleverly started to get around the problem of context—just as commercially available chatbots hope to do. It’s much easier to model an interaction that goes a few layers into the conversational topic if you’re limiting those topics to a specific field.
Developing a machine that can hold almost any conversation with a human interlocutor convincingly might be difficult. It might even be a problem that requires artificial general intelligence to truly solve, rather than the previously-employed approaches of scripted answers or neural networks that associate inputs with responses.
But a machine that can have meaningful interactions that people might value and enjoy could be just around the corner. The Alexa Prize winner is announced in November. The ELIZA effect might mean we will relate to machines sooner than we’d thought.
So, go well, little socialbots. If you ever want to discuss the weather or what the world will be like once you guys take over, I’ll be around. Just don’t start a therapy session.
Image Credit: Shutterstock Continue reading
Posted in Human Robots
#431158 This AI Assistant Helps Demystify ...
In an interview at Singularity University’s Global Summit in San Francisco, Anita Schjøll Brede talked about how artificial intelligence can help make scientific research accessible to anyone working on a complex problem.
Anita Schjøll Brede is the CEO and co-founder of Iris AI, a startup that’s building an artificially intelligent research assistant, which was recently named one of the most innovative AI startups of 2017 by Fast Company. Schjøll Brede is also faculty at Singularity University Denmark and a 2015 alumni of the Global Solutions Program.
“Ultimately, we’re building an AI that can read, understand, and connect the dots,” Schjøll Brede said. “But zooming that back into today, we’re building a tool for R&D, research institutions, and entrepreneurs who have big hairy problems to solve and need to apply research and science to solve them. We’re semi-automating the process of mapping out what you should read to solve the problem or to see what research you need to do to solve the problem.”
Watch the interview for more on Iris AI’s technology and to hear Schjøll Brede’s take on whether AI researchers share a moral responsibility for the systems they build.
Image Credit: foxaon1987 / Shutterstock.com Continue reading
Posted in Human Robots