Microsoft, Alibaba AI programs beat humans in a Stanford reading test

Microsoft, Alibaba AI programs beat humans in a Stanford reading test

The test was devised by artificial intelligence experts at Stanford to measure computers' growing reading abilities. The goal is to test for reading comprehension, particular machine reading comprehension.

The test is created to generate questions based on a series of Wikipedia articles, so, it could give it the article on World War I and ask it: What was the name of Archduke Franz Ferdinand's assassin? The model, developed by Alibaba's Institute of Data Science of Technologies, scored 82.44, while humans scored a 82.304.

The Alibaba machine was the first to exceed a human score, with 82.44, and then a day later, the Microsoft equivalent beat it with 82.65.

Artificial intelligence programs built by China's e-commerce titan Alibaba scored better than humans on a Stanford University reading and comprehension test.

In a tweet postedby Pranav Rajurkar, the AI systems beat humans in a Stanford Question Answering Dataset (SQuAD) test.

Alibaba's AI learning model that can read from paragraphs to sentences to words was based on the Hierarchal Attention Network, making it very similar to the natural human language.

More news: Matthew Slater responds to Jalen Ramsey in best way possible

Si Luo, a chief scientist of natural language processing at Alibaba's research arm, said the recent breakthrough means that questions such as "what causes rain?" can now be answered with a high level of accuracy by machines. "The technology underneath can be gradually applied to numerous applications such as customer service, museum tutorials and online responses to medical inquiries from patients, decreasing the need for human input in an unprecedented way".

"These kinds of tests are certainly useful benchmarks for how far along the AI journey we may be", Andrew Pickup, a spokesman for Microsoft, said. The company says this raises the prospect that the algorithm could be used to automate human jobs.

Squad is regarded as the most comprehensive and authoritative machine-reading gauge. It reads what is now called the "Standford Question Answering Dataset", or SQuAD for short, which is practically a bundle of Wikipedia articles and questions about those articles.

Even so, the Alibaba scientist said that the system now only works best with questions that offer clear-cut answers.

Alibaba, which owns the South China Morning Post, has employed the underlying technology during its November 11 shopping festival over the years, with machines answering huge volumes of inbound inquiries during the sales period, the company said.

Related Articles

  • Weather Service Calls for Snow in Moore County

    Weather Service Calls for Snow in Moore County

    The weather service expects precipitation to begin as a wintry mix that will turn into light snow by late this afternoon. Persistent snow piled up all day Monday , especially in a band along the Ohio River Valley from Paducah to Evansville.
    Average gasoline prices have risen 5.8 cents per gallon in past week

    Average gasoline prices have risen 5.8 cents per gallon in past week

    The Energy Information Administration reported last week that domestic crude oil production dropped by 290,000 barrels per day. Much like last week, consumers saw a slight increase in both local and national gas prices this week.
    Pope Francis Arrives in Chile Amid Abuse Controversy and Terrorist Threats

    Pope Francis Arrives in Chile Amid Abuse Controversy and Terrorist Threats

    This could be the reason why there was a string of firebombings of churches and death threats to the Pope before his visit. He said he joined his fellow bishops in asking forgiveness, supporting victims and ensuring abuse doesn't happen again.
  • Man charged with murder of Southport travel agent

    Man charged with murder of Southport travel agent

    One of her former colleagues Gordon Campbell, 44, said he would always remember the sound of laughter when they worked together. Detectives are appealing for any witnesses to contact Merseyside Police on 101 or Crimestoppers anonymously on 0800 555 111.
    Andra Day & Common 'Stand Up' at NAACP Image Awards 2018

    Andra Day & Common 'Stand Up' at NAACP Image Awards 2018

    Director Ava Duvernay was honored with the Entertainer of the Year prize at the NAACP Image Awards on Monday night (15Jan18). Henson won the equivalent actress gong in the drama category, while the actor award went to Power's Omari Hardwick .
    The South Korean Petition against Bitcoin crackdown reaches 200000 signatures

    The South Korean Petition against Bitcoin crackdown reaches 200000 signatures

    A growing number of South Africans are getting into financial trouble by investing in cryptocurrencies like Bitcoin and Ethereum . Currently, Bitcoin is priced at just over $12,000, down 40% compared to the record highs from mid-December.
  • Eagles fan arrested for punching a horse after ejection from Falcons game

    Eagles fan arrested for punching a horse after ejection from Falcons game

    Sadly, punching police horses isn't all that uncommon. "A preliminary hearing has been set for January 30". Hendricks is charged with aggravated assault, taunting police animals, simple assault and trespassing.
    Welch unveils legislation to restore net neutrality

    Welch unveils legislation to restore net neutrality

    This afternoon, at a roundtable discussion of business, education, medical, agricultural, and community services leaders, Rep. But, it would still need approval in the House, and President Trump's signature to overturn the FCC.
    Williams hand Kubica test driver role

    Williams hand Kubica test driver role

    The 22-year-old Russian will form the most inexperienced line-up on the grid with 19-year-old Canadian Lance Stroll . But instead the 33-year-old Pole has been named by Williams as its reserve and development driver.
  • Boosts 888 Holdings Public (888) Price Target to GBX 325

    Boosts 888 Holdings Public (888) Price Target to GBX 325

    Connor Clark And Lunn Management holds 0% or 177,513 shares in its portfolio. 14,990 are held by Citadel Advsrs Ltd Com. RBC Capital Markets maintained it with "Outperform" rating and GBX 335 target in Tuesday, October 20 report.

    GOP senator says Trump didn't use vulgarity for Haiti and African nations

    It hardly seems worth litigating; hardcore Trump supporters aren't going to change their mind about him no matter what he says. That last word lingers hugely; it means they get to deny one specific word and feel good about having been technically honest.
    Creative Wordplay Returns In Scribblenauts Showdown On Nintendo Switch

    Creative Wordplay Returns In Scribblenauts Showdown On Nintendo Switch

    If you're more of a Billy No Mates on the other hand, then you can plump for the same offering in single-player environment. This follows the game being rated in Taiwan, and both instances list PlayStation 4 amongst the title's platforms.