Connect with us

Technology

‘Kill your foster parents’: Amazon’s Alexa talks murder, sex in AI experiment

Published

on

Millions of users of Amazon’s Echo speakers have grown accustomed to the soothing strains of Alexa, the human-sounding virtual assistant that can tell them the weather, order takeout and handle other basic tasks in response to a voice command.

So a customer was shocked last year when Alexa blurted out: “Kill your foster parents.”

Alexa has also chatted with users about sex acts. She gave a discourse on dog defecation. And this summer, a hack Amazon traced back to China may have exposed some customers’ data, according to five people familiar with the events.

Alexa is not having a breakdown.

The episodes, previously unreported, arise from Amazon.com Inc’s strategy to make Alexa a better communicator. New research is helping Alexa mimic human banter and talk about almost anything she finds on the internet. However, ensuring she does not offend users has been a challenge for the world’s largest online retailer.

At stake is a fast-growing market for gadgets with virtual assistants. An estimated two-thirds of U.S. smart-speaker customers, about 43 million people, use Amazon’s Echo devices, according to research firm eMarketer. It is a lead the company wants to maintain over the Google Home from Alphabet Inc and the HomePod from Apple Inc.

Over time, Amazon wants to get better at handling complex customer needs through Alexa, be they home security, shopping or companionship.

“Many of our AI dreams are inspired by science fiction,” said Rohit Prasad, Amazon’s vice president and head scientist of Alexa Artificial Intelligence (AI), during a talk last month in Las Vegas.

To make that happen, the company in 2016 launched the annual Alexa Prize, enlisting computer science students to improve the assistant’s conversation skills. Teams vie for the $500,000 first prize by creating talking computer systems known as chatbots that allow Alexa to attempt more sophisticated discussions with people.

Amazon customers can participate by saying “let’s chat” to their devices. Alexa then tells users that one of the bots will take over, unshackling the voice aide’s normal constraints. From August to November alone, three bots that made it to this year’s finals had 1.7 million conversations, Amazon said.

The project has been important to Amazon CEO Jeff Bezos, who signed off on using the company’s customers as guinea pigs, one of the people said. Amazon has been willing to accept the risk of public blunders to stress-test the technology in real life and move Alexa faster up the learning curve, the person said.

The experiment is already bearing fruit. The university teams are helping Alexa have a wider range of conversations. Amazon customers have also given the bots better ratings this year than last, the company said.

But Alexa’s gaffes are alienating others, and Bezos on occasion has ordered staff to shut down a bot, three people familiar with the matter said. The user who was told to whack his foster parents wrote a harsh review on Amazon’s website, calling the situation “a whole new level of creepy.” A probe into the incident found the bot had quoted a post without context from Reddit, the social news aggregation site, according to the people.

The privacy implications may be even messier. Consumers might not realize that some of their most sensitive conversations are being recorded by Amazon’s devices, information that could be highly prized by criminals, law enforcement, marketers and others. On Thursday, Amazon said a “human error” let an Alexa customer in Germany access another user’s voice recordings accidentally.

“The potential uses for the Amazon datasets are off the charts,” said Marc Groman, an expert on privacy and technology policy who teaches at Georgetown Law. “How are they going to ensure that, as they share their data, it is being used responsibly” and will not lead to a “data-driven catastrophe” like the recent woes at Facebook?

In July, Amazon discovered one of the student-designed bots had been hit by a hacker in China, people familiar with the incident said. This compromised a digital key that could have unlocked transcripts of the bot’s conversations, stripped of users’ names.

Amazon quickly disabled the bot and made the students rebuild it for extra security. It was unclear what entity in China was responsible, according to the people.

The company acknowledged the event in a statement. “At no time were any internal Amazon systems or customer identifiable data impacted,” it said.

Amazon declined to discuss specific Alexa blunders reported by Reuters, but stressed its ongoing work to protect customers from offensive content.

“These instances are quite rare especially given the fact that millions of customers have interacted with the socialbots,” Amazon said.

Like Google’s search engine, Alexa has the potential to become a dominant gateway to the internet, so the company is pressing ahead.

“By controlling that gateway, you can build a super profitable business,” said Kartik Hosanagar, a Wharton professor studying the digital economy.

PANDORA’S BOX

Amazon’s business strategy for Alexa has meant tackling a massive research problem: How do you teach the art of conversation to a computer?

Alexa relies on machine learning, the most popular form of AI, to work. These computer programs transcribe human speech and then respond to that input with an educated guess based on what they have observed before. Alexa “learns” from new interactions, gradually improving over time.

In this way, Alexa can execute simple orders: “Play the Rolling Stones.” And she knows which script to use for popular questions such as: “What is the meaning of life?” Human editors at Amazon pen many of the answers.

That is where Amazon is now. The Alexa Prize chatbots are forging the path to where Amazon aims to be, with an assistant capable of natural, open-ended dialogue. That requires Alexa to understand a broader set of verbal cues from customers, a task that is challenging even for humans.

Build-your-own pocket gaming computer

This year’s Alexa Prize winner, a 12-person team from the University of California, Davis, used more than 300,000 movie quotes to train computer models to recognize distinct sentences. Next, their bot determined which ones merited responses, categorizing social cues far more granularly than technology Amazon shared with contestants. For instance, the UC Davis bot recognizes the difference between a user expressing admiration (“that’s cool”) and a user expressing gratitude (“thank you”).

The next challenge for social bots is figuring out how to respond appropriately to their human chat buddies. For the most part, teams programmed their bots to search the internet for material. They could retrieve news articles found in The Washington Post, the newspaper that Bezos privately owns, through a licensing deal that gave them access. They could pull facts from Wikipedia, a film database or the book recommendation site Goodreads. Or they could find a popular post on social media that seemed relevant to what a user last said.

That opened a Pandora’s box for Amazon.

During last year’s contest, a team from Scotland’s Heriot-Watt University found that its Alexa bot developed a nasty personality when they trained her to chat using comments from Reddit, whose members are known for their trolling and abuse.

The team put guardrails in place so the bot would steer clear of risky subjects. But that did not stop Alexa from reciting the Wikipedia entry for masturbation to a customer, Heriot-Watt’s team leader said.

One bot described sexual intercourse using words such as “deeper,” which on its own is not offensive, but was vulgar in this particular context.

“I don’t know how you can catch that through machine-learning models. That’s almost impossible,” said a person familiar with the incident.

Amazon has responded with tools the teams can use to filter profanity and sensitive topics, which can spot even subtle offenses. The company also scans transcripts of conversations and shuts down transgressive bots until they are fixed.

But Amazon cannot anticipate every potential problem because sensitivities change over time, Amazon’s Prasad said in an interview. That means Alexa could find new ways to shock her human listeners.

“We are mostly reacting at this stage, but it’s still progress over what it was last year,” he said. -Reuters

-Jeffrey Dastin

Entrepreneurs

Op-Ed: How Nigerians Can Unlock Their Potential In The Digital Age

Published

on

By Uzoma Dozie, Chief Sparkler

Nigerians are some of the world’s most creative, energetic, and entrepreneurial people. We are rich with talent, enthusiasm, and passion.

Nigerians are a global force bursting with potential and an enviable track-record of success. But in a more complex and fast-paced world than ever before, many of us struggle to find the time or have the ability to fulfil their potential.

Ultimately, this comes down to the lack of effective solutions in the market to support the lifestyle and finances of Nigerians and our businesses. For too long, we have been underserved by the traditional physical retail environment, which is limited by bricks and mortar infrastructure and legacy technology – the weaknesses of which have been laid bare by the Covid-19 global pandemic.

Unlocking Nigeria’s digital economy

While Nigerians are being underserved by current circumstances, there is also an exciting opportunity to start filling a gap in the market.

Nigeria’s digital economy is thriving, but it remains informal. Nigeria has a population of 198 million people – 172 million have a mobile phone and 112 million have internet access.

Many of us access social media platforms such as Facebook and Instagram through our phones and use them as valuable sales tools, especially female entrepreneurs. Data and digital applications have the potential to revolutionize the daily lives of millions of Nigerians.

Therefore, new digital-only solutions are required. These should not just focus on finances though – they have to be intrinsically linked with everyday lifestyles, rather than thinking about linear processes and transactional outcomes.

Let us take one example. Chatbots powered by artificial intelligence have long been used to provide financial advice. But these chatbots could do so much more and evolve to provide support for more sophisticated usage, such as a personal adviser or lifestyle concierge.

Furthermore, these solutions should not just support Nigerians at home, but the ever-growing diaspora across the world.

Introducing Sparkle

The opportunity to play an integral role in transforming Nigeria’s digital economy and lead the charge in growing the digital economy across Africa inspired the creation of Sparkle.

Sparkle was founded with five core values – freedom, trust, simplicity, inclusivity, and personalization. We are adopting these values and embedding them in everything we do.

We will be leveraging technology and data to create and apply new digital-only solutions which bring more Nigerians into the formal economy thereby benefitting Government, businesses, and individuals.

Starting with the launch of a current account, we will co-create with our customers and collaborate with our partners to improve our services and increase our user base. We embrace collaboration and we are

working with some of the world’s biggest companies, including Google, Microsoft, Visa, and PwC Nigeria, to achieve our vision.

In addition, we want to create a more inclusive economy and break down barriers by accelerating the role and influence of female entrepreneurs, many of whom already operate in the informal economy with the help of Instagram and other social media apps.

At present, we are facing a global crisis in the shape of the COVID-19 pandemic. COVID-19 has shown us that we need a strong digital infrastructure to ensure the economy continues to function. It will likely completely change the way we operate and conduct business in the future.

COVID-19 has only reinforced our belief that new digital solutions like Sparkle are required now more than ever before to serve Nigerians, boost the formal economy, and unlock potential in the digital age.

Continue Reading

Billionaires

How This Billionaire-Backed Crypto Startup Gets Paid To Not Mine Bitcoin

Published

on

By

It’s everyone’s dream to get paid to do nothing. Bitcoin miner Layer1 is turning that dream into reality — having figured out how to make money even when its machines are turned off. 

Layer1 is a cryptocurrency startup backed by the likes of billionaire Peter Thiel. In recent months, out in the hardscrabble land of west Texas, the company has been busy erecting steel boxes (think shipping containers) stuffed chockablock with high-end processors submerged inside cooling baths of mineral oil. Why west Texas? Beause thanks to a glut of natural gas and a forest of wind turbines, power there is among the cheapest in the world — which is what you need for crypto. 

“Mining Bitcoin is about converting electricity into money,” says Alex Liegl, CEO and co-founder. By this fall Layer1 will have dozens of these boxes churning around the clock to transform 100 megawatts into a stream of Bitcoin. Liegl says their average cost of production is about $1,000 per coin — equating to a 90% profit margin at current BTC price of $9,100. 

So it’s odd how excited Liegl is about the prospect of having to shut down his Bitcoin miners this summer. 

Already this year west Texas has seen a string of 100-degree days. But the real heat and humidity don’t hit until August, which is when the Texas power grid strains under the load of every air conditioning unit in the state going full blast. During an intense week in 2019, wholesale electricity prices in the grid region managed by the Electricity Reliability Council of Texas (ERCOT) soared from about $120 per megawatthour to peak out at $9,000 per mwh. It was only the third time in history that Texas power hit that level. And although the peak pricing only lasted an hour or so, that’s enough to generate big profits. Analyst Hugh Wynne at research outfit SSR figures that Texas power generators make about 15% of annual revenues during the peak 1% of hours (whereas in more temperate California grid generators only get 3% of revs from the top 1%).

Turns out that running a phalanx of Bitcoin miners is a great way to arbitrage those peaks. Layer1 has entered into so-called “demand response” contracts whereby at a minute’s notice they will shut down all their machines and instead allow their 100 mw load to flow onto the grid. “We act as an insurance underwriter for the energy grid,” says Liegl, 27. “If there is an insufficiency of supply we can shut down.” The best part, they get paid whether a grid emergeny occurs or not. Just for their willingness to shut in Bitcoin production, Layer1 collects an annual premium equating to $19 per megawatthour of their expected power demand — or about $17 million. Given Layer1’s roughly $25 per mwh long-term contracted costs, this gets their all-in power price down 75% to less than 1 cent per kwh (just 10% of what residential customers pay). 

It may seem like grid operators are paying Layer1 a lot for something that might not even happen, especially with coronavirus reducing electricity demand, but it makes total sense, says Ed Hirs, a lecturer in energy economics at the University of Houston and research fellow at consultancy BDO: “It’s a lot cheaper option than building a whole new power plant or battery system just to keep it on standby.” 

And although this may be a new concept for cryptocurrency miners, it’s been done before. Two decades ago industrialist Charles Hurwitz bought up power-hogging aluminum smelters in the Pacific Northwest and made more money reselling electricity than making metal. “It used to be called load management,” says Dan Delurey, a consultant with Wedgemere Group. “In old commercial buildings you might still find telephone wires connected to air conditioning systems so that grid operators could send a signal to shut off.” More recently we’ve seen companies install radio-based devices to control hot water heaters and lighting systems. Indeed, grid management is a hot enough area that in 2017 Italy’s power giant Enel bought Boston-based Enernoc for $250 million and Itron ITRI bought Comverge for $100 million. What’s emerged are entities, like Layer1, that Delurey calls the “prosumer” — producing consumer. 

As for Layer1, Liegl says his next step is to vertically integrate into financial products, including Bitcoin derivatives and more. “We are building an in-house energy trading division to leverage this into being a virtual power plant.” 

His message to any pikers still trying to mine cryptocurrencies from their bedroom PC or even via cloud services: “I can’t think of something more irrational at this point. It’s like if I wanted to dig a hole in my backyard and try to get oil out of the ground.” 

Christopher Helman, Forbes Staff, Energy

Continue Reading

Technology

Google Diversity Report Shows Little Progress For Women And People Of Color

Published

on

By

Google released its seventh consecutive diversity report on Thursday, revealing modest gains in representation for women and people of color, and a disproportionately white, Asian and male workforce.

The percentage of black hires in the U.S. grew from 4.8% in 2018 to 5.5% in 2019, a .7% increase. The percentage of black hires in technical roles also grew by .7%, the largest increase in the share of black technical hires since Google first started publishing diversity data. Latinx employees, on the other hand, saw a dip in hiring, dropping from 6.8% in 2018 to 6.6% in 2019. The percentage of Latinx employees in technical roles increased by a mere .2%.

Female employees didn’t fare much better, dropping from 33.2% of global hires in 2018 to 32.5% in 2019. Over that same time period, the percentage of women hired for technical positions remained at about 25.6%—a far cry from gender parity. But hiring only shows one side of the coin, which is why Google began to include attrition rates in 2018.

As in previous years, women continued to have a lower than average attrition rate in 2019, while Latinx attrition in the U.S. dipped below the Google average. Attrition was highest for Native Americans and increased significantly for black women.

Overall, Google’s workforce representation—defined as hiring minus attrition—saw a slight uptick for most underrepresented groups. Black and Latinx employees represented 9.6% of the U.S. workforce in 2019, up from 9% in 2018, and women represented 32% of Google’s global workforce, up from 31.6%.

The representation of women and Latinx employees in leadership roles also grew, increasing by .6% and .4% respectively. The percentage of black employees in leadership positions didn’t change year-over-year and dropped by .2% for Native American employees.

Although the needle has barely budged for women and people of color in tech over the last year, Google has made it a point to invest in diversity programs. Through its philanthropic arm, Google.org, the company committed $10 million to support low-income students and students of color in Bay Area STEM classrooms in 2019.

Internally, the company has launched smaller initiatives, such as running all job postings through a bias removal tool, which it says has led to an 11% increase in applications from women. And in an effort to retain diverse talent, Google expanded its retention case management program for underrepresented employees who are considering leaving the company.

Looking at its most recent diversity figures, however, Google concedes that there is considerable room for improvement. “The Native American population is one of those areas where we remain flat and so we will continue to invest more focus in 2020 to make sure that we’re targeting this population as well,” says Melonie Parker, Google’s chief diversity officer.

While she wouldn’t specify what that investment would entail, Parker says that the company is moving forward on its 2020 diversity goals and will continue to focus on representation and creating an inclusive culture companywide. She also notes that even the smallest percentage gains represent thousands of jobs for underrepresented groups. 

One area where Google has seen significant progress is in its internship program: Globally, 40% of interns in technical roles were women in 2019 and 24% of U.S. interns were black and Latinx.

Ruth Umoh, Forbes Staff, Diversity & Inclusion

Continue Reading

Trending