Stories
Slash Boxes
Comments

SoylentNews is people

posted by hubie on Sunday March 12 2023, @04:50PM   Printer-friendly
from the keeping-up-with-the-joneses dept.

https://arstechnica.com/information-technology/2023/03/wikipedia-ai-truth-duckduckgo-hopes-so-with-new-answerbot/

Not to be left out of the rush to integrate generative AI into search, on Wednesday DuckDuckGo announced DuckAssist, an AI-powered factual summary service powered by technology from Anthropic and OpenAI. It is available for free today as a wide beta test for users of DuckDuckGo's browser extensions and browsing apps. Being powered by an AI model, the company admits that DuckAssist might make stuff up but hopes it will happen rarely.

Here's how it works: If a DuckDuckGo user searches a question that can be answered by Wikipedia, DuckAssist may appear and use AI natural language technology to generate a brief summary of what it finds in Wikipedia, with source links listed below. The summary appears above DuckDuckGo's regular search results in a special box.

[...] Update (March 9, 2023): We spoke with a representative of DuckDuckGo and they said they're using OpenAI's GPT-3.5 and Anthropic's Claude as LLMs. "We're experimenting with OpenAI's recently announced Turbo model, too," they said.

Related:
Robots Let ChatGPT Touch the Real World Thanks to Microsoft (Article has a bunch of other SoylentNews related links as well.)


Original Submission

Related Stories

Robots Let ChatGPT Touch the Real World Thanks to Microsoft 15 comments

https://arstechnica.com/information-technology/2023/02/robots-let-chatgpt-touch-the-real-world-thanks-to-microsoft/

Last week, Microsoft researchers announced an experimental framework to control robots and drones using the language abilities of ChatGPT, a popular AI language model created by OpenAI. Using natural language commands, ChatGPT can write special code that controls robot movements. A human then views the results and adjusts as necessary until the task gets completed successfully.

The research arrived in a paper titled "ChatGPT for Robotics: Design Principles and Model Abilities," authored by Sai Vemprala, Rogerio Bonatti, Arthur Bucker, and Ashish Kapoor of the Microsoft Autonomous Systems and Robotics Group.

In a demonstration video, Microsoft shows robots—apparently controlled by code written by ChatGPT while following human instructions—using a robot arm to arrange blocks into a Microsoft logo, flying a drone to inspect the contents of a shelf, or finding objects using a robot with vision capabilities.

To get ChatGPT to interface with robotics, the researchers taught ChatGPT a custom robotics API. When given instructions like "pick up the ball," ChatGPT can generate robotics control code just as it would write a poem or complete an essay. After a human inspects and edits the code for accuracy and safety, the human operator can execute the task and evaluate its performance.

In this way, ChatGPT accelerates robotic control programming, but it's not an autonomous system. "We emphasize that the use of ChatGPT for robotics is not a fully automated process," reads the paper, "but rather acts as a tool to augment human capacity."

You Can Now Run a GPT-3-Level AI Model on Your Laptop, Phone, and Raspberry Pi 30 comments

https://arstechnica.com/information-technology/2023/03/you-can-now-run-a-gpt-3-level-ai-model-on-your-laptop-phone-and-raspberry-pi/

Things are moving at lightning speed in AI Land. On Friday, a software developer named Georgi Gerganov created a tool called "llama.cpp" that can run Meta's new GPT-3-class AI large language model, LLaMA, locally on a Mac laptop. Soon thereafter, people worked out how to run LLaMA on Windows as well. Then someone showed it running on a Pixel 6 phone, and next came a Raspberry Pi (albeit running very slowly).

If this keeps up, we may be looking at a pocket-sized ChatGPT competitor before we know it.
[...]
For example, here's a list of notable LLaMA-related events based on a timeline Willison laid out in a Hacker News comment:

Related:
DuckDuckGo's New Wikipedia Summary Bot: "We Fully Expect It to Make Mistakes"
Robots Let ChatGPT Touch the Real World Thanks to Microsoft (Article has a bunch of other SoylentNews related links as well.)
Netflix Stirs Fears by Using AI-Assisted Background Art in Short Anime Film
Paper: Stable Diffusion "Memorizes" Some Images, Sparking Privacy Concerns
The EU's AI Act Could Have a Chilling Effect on Open Source Efforts, Experts Warn
Pixel Art Comes to Life: Fan Upgrades Classic MS-DOS Games With AI


Original Submission

Baidu Shares Fall After Ernie AI Chatbot Demo Disappoints 7 comments

https://arstechnica.com/information-technology/2023/03/chinese-search-giant-launches-ai-chatbot-with-prerecorded-demo/

Shares of Baidu fell as much as 10 percent on Thursday after the web search company showed only a pre-recorded video of its AI chatbot Ernie in the first public release of China's answer to ChatGPT.

The Beijing-based tech company has claimed Ernie will remake its business and for weeks talked up plans to incorporate generative artificial intelligence into its search engine and other products.

But on Thursday, millions of people tuning in to the event were left with little idea of whether Baidu's chatbot could compete with ChatGPT.
[...]
"We can only explore by ourselves. Training ChatGPT took OpenAI more than a year, and it took them another year to tune GPT-4," said one Baidu employee. "It means we're two years behind."

Baidu did not immediately respond to a request for comment.

Related:
The AI Hype Bubble is the New Crypto Hype Bubble
DuckDuckGo's New Wikipedia Summary Bot: "We Fully Expect It to Make Mistakes"
LLM ChatGPT Might Change the World, but Not in a Good Way
Alphabet Stock Price Drops After Google Bard Launch Blunder
OpenAI and Microsoft Announce Extended, Multi-Billion-Dollar Partnership


Original Submission

This discussion was created by hubie (1068) for logged-in users only, but now has been archived. No new comments can be posted.
Display Options Threshold/Breakthrough Mark All as Read Mark All as Unread
The Fine Print: The following comments are owned by whoever posted them. We are not responsible for them in any way.
(1)
  • (Score: 5, Insightful) by Rosco P. Coltrane on Sunday March 12 2023, @06:45PM (10 children)

    by Rosco P. Coltrane (4757) on Sunday March 12 2023, @06:45PM (#1295800)

    People my age remember search engines where one could enter almost regex-like queries. Well, not regexes strictly-speaking, but if you knew you wanted a PDF file from a site in Germany with "manual" in the filename and not containing "video" for instance, you could do something like "file:manual.*.pdf +site:de -video". Maybe not that exact syntax, and it varied from search engine to search engine, but if you knew exactly what you were looking for, you could find it quickly and precisely with the right search constraints.

    If you do that in Google now, it'll return random hits from the highest paying advertisers, followed by stuff you precisely don't want from servers in countries you're explictely not interested in and containing stuff you told it to ignore, because the search has gone "soft" many years ago: it's not a strict set of requirement for your search query anymore, it's kind of a vague guideline for Google to serve you stuff that will make them the most money while still looking vaguely relevant to your query.

    Now we're about to fight deluded AIs that not only don't understand what you want, but try to interpret it - because the assumption of those who push those things is that the user is a dumbass who doesn't really know what they're looking for - in the best of case, and reply plain lie or hallucinate truthiness in the worst of case.

    Not only won't you find what you want like before, but on top of that, you're going to try to convince the AI that it doesn't know better than you, then double-check what it says in a regular search engine to be sure it's not telling you BS.

    I don't look forward to interacting with those AIs in my daily searches, exactly the same way I dread dealing with any AI that deal with any ticket I open with any company or admininstration to deal with a very precise problem that would take 5 minutes for a real human to understand, but you can't resolve because the support guy is a now a dumb machine that guestimates the best boilerplate FAQ entry to send you in response to your query.

    Sheesh, I don't know why the future of computing keeps turning into a more and more useless and depressing present... I'm sure AI will become more useful than humans, and probably sentient at some point, but before we reach that point, we have at least a solid decade of information idiocy in front of us...

    • (Score: 2, Informative) by Anonymous Coward on Sunday March 12 2023, @07:23PM

      by Anonymous Coward on Sunday March 12 2023, @07:23PM (#1295803)

      People my age remember when the only search engine was the card catalog at the city library.

    • (Score: 1, Touché) by Anonymous Coward on Sunday March 12 2023, @08:38PM (3 children)

      by Anonymous Coward on Sunday March 12 2023, @08:38PM (#1295810)

      I don't know why the future of computing keeps turning into a more and more useless and depressing present

      You don't know why?! Do you want to take a guess?

      • (Score: 0) by Anonymous Coward on Sunday March 12 2023, @09:29PM (1 child)

        by Anonymous Coward on Sunday March 12 2023, @09:29PM (#1295817)

        I would state the top reasons are:
        -Hiring less capable, less experienced, poor fit candidates for the position just to check the box labeled Diversity and Inclusion?
        -Hiring art degree majors with no CS experience and no concept of good UX to design the UI
        -No criticism of bad ideas is allowed in fear of cancel culture, aka bullying-witchhunt-opression-mobocracy. When someone comes up with a ridiculous idea, no one can say anymore "This is a dumb idea" anymore out of fear of causing a torrent of "I am literally shaking right now" tears to rain down. Basically bad ideas cannot be questioned anymore, especially when they come from the D&I people.
        -All good ideas and designs from the past are being thrown out because of "Ewww... some greedy boomer designed that in 2006, we have to throw that out."

        • (Score: 0) by Anonymous Coward on Sunday March 12 2023, @10:18PM

          by Anonymous Coward on Sunday March 12 2023, @10:18PM (#1295821)

          BZZZZT.

          So these companies that fire 10,000 employees at a time are squeamish about hurting Millenial's feelz and get stepped all over? Poor companies, I feelz sad now :(

      • (Score: 1) by khallow on Monday March 13 2023, @03:48PM

        by khallow (3766) Subscriber Badge on Monday March 13 2023, @03:48PM (#1295912) Journal
        It's the McDonalds ice cream effect. They can't change the price, but they can shave cents off the cost of the ice cream cone by reducing quality in relatively hidden and incremental ways (like smaller portion sizes, lower quality cone wafers, and more air in the ice cream mix). We can't point to any obvious point in the process where it got worse, but if you compare quality of the product years apart, you can see that something happened.

        Here, they're probably doing less extensive and focused searching behind the scenes to reduce costs. It's also possible that a decade of fighting search engine optimization has crippled the search functionality too.
    • (Score: 3, Interesting) by istartedi on Sunday March 12 2023, @08:53PM (1 child)

      by istartedi (123) on Sunday March 12 2023, @08:53PM (#1295814) Journal

      Just a couple days ago people were commenting about the "big snowflakes" in Iowa. I recalled that my Dad always called them "goose feathers". I was wondering about the origin of that--whether it was a regionalism or not. Try as I might, I couldn't find an easy way to build that kind of query and avoid all the hits that were about literal geese and feathers, or snow geese. It totally ignored "goose feathers" in quotes, and it wasn't even trying to shove ads that much. It just sucked. In the old days it seems like I might have come across some essay by an English professor explaining the fascinating etymology of the expression and/or the regions in which it's used. Or for that matter, a confirmation from a meteorologist of my suspicion that such snow indicates you're near the rain-snow boundary. For all I know that kind of content is still out there, stranded beyond the reach of our currently dumbed-down search.

      --
      Appended to the end of comments you post. Max: 120 chars.
    • (Score: 5, Insightful) by BlueCoffee on Sunday March 12 2023, @09:12PM (1 child)

      by BlueCoffee (18257) on Sunday March 12 2023, @09:12PM (#1295815)

      I hear ya! Search engine results have been crap for a long time because the business suits think that returning 1900 general results is better than returning 200 specific results for various reasons:
      -terms are ORed instead of ANDed so more results are returned
      -they silently substitute words to return more results
      -when you use +, -, and quotes there is a high probability those switches will get ignored.

      • (Score: 2) by cmdrklarg on Tuesday March 14 2023, @10:04PM

        by cmdrklarg (5048) Subscriber Badge on Tuesday March 14 2023, @10:04PM (#1296166)

        You're already at +5 so I can't add more. Amazon does this very thing, making looking for anything specific there an enormous waste of time.

        Plenty of Fish's phone app did this shit too. Search for female, 38-57 years old, non-religious, within 25 miles. "Oh, you needed more results so we changed your search parameters". Assholes.

        --
        The world is full of kings and queens who blind your eyes and steal your dreams.
    • (Score: 0) by Anonymous Coward on Monday March 13 2023, @06:39AM

      by Anonymous Coward on Monday March 13 2023, @06:39AM (#1295860)
      Possible bad scenario - search engines find you lots of AI created webpages with inaccurate info.
  • (Score: 5, Insightful) by Snotnose on Sunday March 12 2023, @07:03PM (6 children)

    by Snotnose (1623) on Sunday March 12 2023, @07:03PM (#1295801)

    A day or three ago I read an article about this very thing. In the beginning websites want users, so they cater to them. Once they have a bunch of users they want advertisers, so they cater to them. Once they have a bunch of those they figure both the users and the advertisers are locked into the platform and extract every last dime they can, screwing over both users and advertisers.

    Back in the mid 90s Alta Vista was the best search engine (for me, YMMV). Then they announced they were going to sell search results, top listed was who paid the most, 2nd was who paid second most, etc. The users fled in droves before they could even implement it. For a while we all kept looking for a decent search engine until Google showed up.

    I don't know why Google is still so popular. As the above poster said, I used to use type:, site: etc tags a lot. They typically no longer work. I switch to DDG several years ago and only use Google when DDG won't show a result I want. Two times out of three Google also doesn't show a useful result either.

    --
    When the dust settled America realized it was saved by a porn star.
    • (Score: 4, Informative) by gznork26 on Sunday March 12 2023, @07:23PM

      by gznork26 (1159) on Sunday March 12 2023, @07:23PM (#1295804) Homepage Journal

      That sounds like one of Cory Doctorow's recent articles. Well worth visiting Pluralistic for on a regular basis.

      --
      Khipu were Turing complete.
    • (Score: 4, Informative) by kazzie on Sunday March 12 2023, @07:38PM (3 children)

      by kazzie (5309) Subscriber Badge on Sunday March 12 2023, @07:38PM (#1295805)

      Possibly in the Guardian? They published an article quoting Cory Doctorow and his theory of enshittification yesterday:

      https://www.theguardian.com/commentisfree/2023/mar/11/users-advertisers-we-are-all-trapped-in-the-enshittification-of-the-internet [theguardian.com]

      • (Score: 1, Touché) by Anonymous Coward on Sunday March 12 2023, @08:43PM (1 child)

        by Anonymous Coward on Sunday March 12 2023, @08:43PM (#1295812)

        Hey, give us some love. We posted it too [soylentnews.org]!

        • (Score: 2) by kazzie on Wednesday March 15 2023, @06:08AM

          by kazzie (5309) Subscriber Badge on Wednesday March 15 2023, @06:08AM (#1296212)

          That could also be where I knew it from.

          Having re-read that fine article, and the one I linked, both cover the same points that I recalled.

      • (Score: 0) by Anonymous Coward on Monday March 13 2023, @03:50PM

        by Anonymous Coward on Monday March 13 2023, @03:50PM (#1295914)

        Seems like cover for ThEy'Re aLL tHe sAmE logic, as applied to politicians. When one side (and I think we all know which one) tries to subvert democracy and gets caught, then muddying the waters with nonsense about space lasers ensures that all news information is perceived as equally dubious.

    • (Score: 0) by Anonymous Coward on Sunday March 12 2023, @08:41PM

      by Anonymous Coward on Sunday March 12 2023, @08:41PM (#1295811)

      I love it when google leaves out the most information dense keyword in the search and gives you a bunch of random garbage instead. Gee, thanks. Good try google.

(1)