Tag Archives: safe
This is part six of a six-part series on the history of natural language processing.
In February of this year, OpenAI, one of the foremost artificial intelligence labs in the world, announced that a team of researchers had built a powerful new text generator called the Generative Pre-Trained Transformer 2, or GPT-2 for short. The researchers used a reinforcement learning algorithm to train their system on a broad set of natural language processing (NLP) capabilities, including reading comprehension, machine translation, and the ability to generate long strings of coherent text.
But as is often the case with NLP technology, the tool held both great promise and great peril. Researchers and policy makers at the lab were concerned that their system, if widely released, could be exploited by bad actors and misappropriated for “malicious purposes.”
The people of OpenAI, which defines its mission as “discovering and enacting the path to safe artificial general intelligence,” were concerned that GPT-2 could be used to flood the Internet with fake text, thereby degrading an already fragile information ecosystem. For this reason, OpenAI decided that it would not release the full version of GPT-2 to the public or other researchers.
GPT-2 is an example of a technique in NLP called language modeling, whereby the computational system internalizes a statistical blueprint of a text so it’s able to mimic it. Just like the predictive text on your phone—which selects words based on words you’ve used before—GPT-2 can look at a string of text and then predict what the next word is likely to be based on the probabilities inherent in that text.
GPT-2 can be seen as a descendant of the statistical language modeling that the Russian mathematician A. A. Markov developed in the early 20th century (covered in part three of this series).
GPT-2 used cutting-edge machine learning algorithms to do linguistic analysis with over 1.5 million parameters.
What’s different with GPT-2, though, is the scale of the textual data modeled by the system. Whereas Markov analyzed a string of 20,000 letters to create a rudimentary model that could predict the likelihood of the next letter of a text being a consonant or a vowel, GPT-2 used 8 million articles scraped from Reddit to predict what the next word might be within that entire dataset.
And whereas Markov manually trained his model by counting only two parameters—vowels and consonants—GPT-2 used cutting-edge machine learning algorithms to do linguistic analysis with over 1.5 million parameters, burning through huge amounts of computational power in the process.
The results were impressive. In their blog post, OpenAI reported that GPT-2 could generate synthetic text in response to prompts, mimicking whatever style of text it was shown. If you prompt the system with a line of William Blake’s poetry, it can generate a line back in the Romantic poet’s style. If you prompt the system with a cake recipe, you get a newly invented recipe in response.
Perhaps the most compelling feature of GPT-2 is that it can answer questions accurately. For example, when OpenAI researchers asked the system, “Who wrote the book The Origin of Species?”—it responded: “Charles Darwin.” While only able to respond accurately some of the time, the feature does seem to be a limited realization of Gottfried Leibniz’s dream of a language-generating machine that could answer any and all human questions (described in part two of this series).
After observing the power of the new system in practice, OpenAI elected not to release the fully trained model. In the lead up to its release in February, there had been heightened awareness about “deepfakes”—synthetic images and videos, generated via machine learning techniques, in which people do and say things they haven’t really done and said. Researchers at OpenAI worried that GPT-2 could be used to essentially create deepfake text, making it harder for people to trust textual information online.
Responses to this decision varied. On one hand, OpenAI’s caution prompted an overblown reaction in the media, with articles about the “dangerous” technology feeding into the Frankenstein narrative that often surrounds developments in AI.
Others took issue with OpenAI’s self-promotion, with some even suggesting that OpenAI purposefully exaggerated GPT-2s power in order to create hype—while contravening a norm in the AI research community, where labs routinely share data, code, and pre-trained models. As machine learning researcher Zachary Lipton tweeted, “Perhaps what's *most remarkable* about the @OpenAI controversy is how *unremarkable* the technology is. Despite their outsize attention & budget, the research itself is perfectly ordinary—right in the main branch of deep learning NLP research.”
OpenAI stood by its decision to release only a limited version of GPT-2, but has since released larger models for other researchers and the public to experiment with. As yet, there has been no reported case of a widely distributed fake news article generated by the system. But there have been a number of interesting spin-off projects, including GPT-2 poetry and a webpage where you can prompt the system with questions yourself.
Mimicking humans on Reddit, the bots have long conversations about a variety of topics, including conspiracy theories and
Star Wars movies.
There’s even a Reddit group populated entirely with text produced by GPT-2-powered bots. Mimicking humans on Reddit, the bots have long conversations about a variety of topics, including conspiracy theories and Star Wars movies.
This bot-powered conversation may signify the new condition of life online, where language is increasingly created by a combination of human and non-human agents, and where maintaining the distinction between human and non-human, despite our best efforts, is increasingly difficult.
The idea of using rules, mechanisms, and algorithms to generate language has inspired people in many different cultures throughout history. But it’s in the online world that this powerful form of wordcraft may really find its natural milieu—in an environment where the identity of speakers becomes more ambiguous, and perhaps, less relevant. It remains to be seen what the consequences will be for language, communication, and our sense of human identity, which is so bound up with our ability to speak in natural language.
This is the sixth installment of a six-part series on the history of natural language processing. Last week’s post explained how an innocent Microsoft chatbot turned instantly racist on Twitter.
You can also check out our prior series on the untold history of AI. Continue reading
Welcome to the eighth edition of IEEE Spectrum’s Robot Gift Guide!
This year we’re featuring 15 robotic products that we think will make fantastic holiday gifts. As always, we tried to include a broad range of robot types and prices, focusing mostly on items released this year. (A reminder: While we provide links to places where you can buy these items, we’re not endorsing any in particular, and a little bit of research may result in better deals.)
If you need even more robot gift ideas, take a look at our past guides: 2018, 2017, 2016, 2015, 2014, 2013, and 2012. Some of those robots are still great choices and might be way cheaper now than when we first posted about them. And if you have suggestions that you’d like to share, post a comment below to help the rest of us find the perfect robot gift.
What makes robots so compelling is their autonomy, and the Skydio 2 is one of the most autonomous robots we’ve ever seen. It uses an array of cameras to map its environment and avoid obstacles in real-time, making flight safe and effortless and enabling the kinds of shots that would be impossible otherwise. Seriously, this thing is magical, and it’s amazing that you can actually buy one.
UBTECH Jimu MeeBot 2
The Jimu MeeBot 2.0 from UBTECH is a STEM education robot designed to be easy to build and program. It includes six servo motors, a color sensor, and LED lights. An app for iPhone or iPad provides step-by-step 3D instructions, and helps you code different behaviors for the robot. It’s available exclusively from Apple.
iRobot Roomba s9+
We know that $1,400 is a crazy amount of money to spend on a robot vacuum, but the Roomba s9+ is a crazy robot vacuum. As if all of its sensors and mapping intelligence wasn’t enough, it empties itself, which means that you can have your floors vacuumed every single day for a month and you don’t have to even think about it. This is what home robots are supposed to be.
Photo: Piaggio Fast Forward
Nobody likes carrying things, which is why Gita is perfect for everyone with an extra $3,000 lying around. Developed by Piaggio Fast Forward, this autonomous robot will follow you around with a cargo hold full of your most important stuff, and do it in a way guaranteed to attract as much attention as possible.
DJI Mavic Mini
It’s tiny, it’s cheap, and it takes good pictures—what more could you ask for from a drone? And for $400, this is an excellent drone to get if you’re on a budget and comfortable with manual flight. Keep in mind that while the Mavic Mini is small enough that you don’t need to register it with the FAA, you do still need to follow all the same rules and regulations.
LEGO Star Wars Droid Commander
Designed for kids ages 8+, this LEGO set includes more than 1,000 pieces, enough to build three different droids: R2-D2, Gonk Droid, and Mouse Droid. Using a Bluetooth-controlled robotic brick called Move Hub, which connects to the LEGO BOOST Star Wars app, kids can change how the robots behave and solve challenges, learning basic robotics and coding skills.
Robot pets don’t get much more sophisticated (or expensive) than Sony’s Aibo. Strictly speaking, it’s one of the most complex consumer robots you can buy, and Sony continues to add to Aibo’s software. Recent new features include user programmability, and the ability to “feed” it.
$2,900 (free aibone and paw pads until 12/29/2019)
Neato Botvac D4 Connected
The Neato Botvac D4 may not have all of the features of its fancier and more expensive siblings, but it does have the features that you probably care the most about: The ability to make maps of its environment for intelligent cleaning (using lasers!), along with user-defined no-go lines that keep it where you want it. And it cleans quite well, too.
$530 $350 (sale)
Cubelets Curiosity Set
Photo: Modular Robotics
Cubelets are magnetic blocks that you can snap together to make an endless variety of robots with no programming and no wires. The newest set, called Curiosity, is designed for kids ages 4+ and comes with 10 robotic cubes. These include light and distance sensors, motors, and a Bluetooth module, which connects the robot constructions to the Cubelets app.
Photo: Franklin Robotics
Tertill does one simple job: It weeds your garden. It’s waterproof, dirt proof, solar powered, and fully autonomous, meaning that you can leave it out in your garden all summer and just enjoy eating your plants rather than taking care of them.
Root was originally developed by Harvard University as a tool to help kids progressively learn to code. iRobot has taken over Root and is now supporting the curriculum, which starts for kids before they even know how to read and should keep them busy for years afterwards.
Let’s be honest: Nobody is really quite sure what LOVOT is. We can all agree that it’s kinda cute, though. And kinda weird. But cute. Created by Japanese robotics startup Groove X, LOVOT does have a whole bunch of tech packed into its bizarre little body and it will do its best to get you to love it.
RVR is a rugged, versatile, easy to program mobile robot. It’s a development platform designed to be a bridge between educational robots like Sphero and more sophisticated and expensive systems like Misty. It’s mostly affordable, very expandable, and comes from a company with a lot of experience making robots.
“How to Train Your Robot”
Image: Lawrence Hall of Science
Aimed at 4th and 5th graders, “How to Train Your Robot,” written by Blooma Goldberg, Ken Goldberg, and Ashley Chase, and illustrated by Dave Clegg, is a perfect introduction to robotics for kids who want to get started with designing and building robots. But the book isn’t just for beginners: It’s also a fun, inspiring read for kids who are already into robotics and want to go further—it even introduces concepts like computer simulations and deep learning. You can download a free digital copy or request hardcopies here.
MIT Mini Cheetah
Yes, Boston Dynamics’ Spot, now available for lease, is probably the world’s most famous quadruped, but MIT is starting to pump out Mini Cheetahs en masse for researchers, and while we’re not exactly sure how you’d manage to get one of these things short of stealing one directly for MIT, a Mini Cheetah is our fantasy robotics gift this year. Mini Cheetah looks like a ton of fun—it’s portable, highly dynamic, super rugged, and easy to control. We want one!
MIT Biomimetic Robotics Lab
For more tech gift ideas, see also IEEE Spectrum’s annual Gift Guide. Continue reading
There’s no particular reason why knowing how to juggle would be a useful skill for a robot. Despite this, robots are frequently taught how to juggle things. Blind robots can juggle, humanoid robots can juggle, and even drones can juggle. Why? Because juggling is hard, man! You have to think about a bunch of different things at once, and also do a bunch of different things at once, which this particular human at least finds to be overly stressful. While juggling may not stress robots out, it does require carefully coordinated sensing and computing and actuation, which means that it’s as good a task as any (and a more entertaining task than most) for testing the capabilities of your system.
UC Berkeley’s Cassie Cal robot, which consists of two legs and what could be called a torso if you were feeling charitable, has just learned to juggle by bouncing a ball on what would be her head if she had one of those. The idea is that if Cassie can juggle while balancing at the same time, she’ll be better able to do other things that require dynamic multitasking, too. And if that doesn’t work out, she’ll still be able to join the circus.
Cassie’s juggling is assisted by an external motion capture system that tracks the location of the ball, but otherwise everything is autonomous. Cassie is able to juggle the ball by leaning forwards and backwards, left and right, and moving up and down. She does this while maintaining her own balance, which is the whole point of this research—successfully executing two dynamic behaviors that may sometimes be at odds with one another. The end goal here is not to make a better juggling robot, but rather to explore dynamic multitasking, a skill that robots will need in order to be successful in human environments.
This work is from the Hybrid Robotics Lab at UC Berkeley, led by Koushil Sreenath, and is being done by Katherine Poggensee, Albert Li, Daniel Sotsaikich, Bike Zhang, and Prasanth Kotaru.
For a bit more detail, we spoke with Albert Li via email.
Image: UC Berkeley
UC Berkeley’s Cassie Cal getting ready to juggle.
IEEE Spectrum: What would be involved in getting Cassie to juggle without relying on motion capture?
Albert Li: Our motivation for starting off with motion capture was to first address the control challenge of juggling on a biped without worrying about implementing the perception. We actually do have a ball detector working on a camera, which would mean we wouldn’t have to rely on the motion capture system. However, we need to mount the camera in a way that it would provide the best upwards field of view, and we also have develop a reliable estimator. The estimator is particularly important because when the ball gets close enough to the camera, we actually can’t track the ball and have to assume our dynamic models describe its motion accurately enough until it bounces back up.
What keeps Cassie from juggling indefinitely?
There are a few factors that affect how long Cassie can sustain a juggle. While in simulation the paddle exhibits homogeneous properties like its stiffness and damping, in reality every surface has anisotropic contact properties. So, there are parts of the paddle which may be better for juggling than others (and importantly, react differently than modeled). These differences in contact are also exacerbated due to how the paddle is cantilevered when mounted on Cassie. When the ball hits these areas, it leads to a larger than expected error in a juggle. Due to the small size of the paddle, the ball may then just hit the paddle’s edge and end the juggling run. Over a very long run, this is a likely occurrence. Additionally, some large juggling errors could cause Cassie’s feet to slip slightly, which ends up changing the stable standing position over time. Since this version of the controller assumes Cassie is stationary, this change in position eventually leads to poor juggles and failure.
Would Cassie be able to juggle while walking (or hovershoe-ing)?
Walking (and hovershoe-ing) while juggling is a far more challenging problem and is certainly a goal for future research. Some of these challenges include getting the paddle to precise poses to juggle the ball while also moving to avoid any destabilizing effects of stepping incorrectly. The number of juggles per step of walking could also vary and make the mathematics of the problem more challenging. The controller goal is also more involved. While the current goal of the juggling controller is to juggle the ball to a static apex position, with a walking juggling controller, we may instead want to hit the ball forwards and also walk forwards to bounce it, juggle the ball along a particular path, etc. Solving such challenges would be the main thrusts of the follow-up research.
Can you give an example of a practical task that would be made possible by using a controller like this?
Studying juggling means studying contact behavior and leveraging our models of it to achieve a known objective. Juggling could also be used to study predictable post-contact flight behavior. Consider the scenario where a robot is attempting to make a catch, but fails, letting the ball to bounce off of its hand, and then recovering the catch. This behavior could also be intentional: It is often easier to first execute a bounce to direct the target and then perform a subsequent action. For example, volleyball players could in principle directly hit a spiked ball back, but almost always bump the ball back up and then return it.
Even beyond this motivating example, the kinds of models we employ to get juggling working are more generally applicable to any task that involves contact, which could include tasks besides bouncing like sliding and rolling. For example, clearing space on a desk by pushing objects to the side may be preferable than individually manipulating each and every object on it.
You mention collaborative juggling or juggling multiple balls—is that something you’ve tried yet? Can you talk a bit more about what you’re working on next?
We haven’t yet started working on collaborative or multi-ball juggling, but that’s also a goal for future work. Juggling multiple balls statically is probably the most reasonable next goal, but presents additional challenges. For instance, you have to encode a notion of juggling urgency (if the second ball isn’t hit hard enough, you have less time to get the first ball up before you get back to the second one).
On the other hand, collaborative human-robot juggling requires a more advanced decision-making framework. To get robust multi-agent juggling, the robot will need to employ some sort of probabilistic model of the expected human behavior (are they likely to move somewhere? Are they trying to catch the ball high or low? Is it safe to hit the ball back?). In general, developing such human models is difficult since humans are fairly unpredictable and often don’t exhibit rational behavior. This will be a focus of future work.
[ Hybrid Robotics Lab ] Continue reading