Please create an account to participate in the Slashdot moderation system

 



Forgot your password?
typodupeerror
×

Submission + - 'Dream glove' boosts creativity during sleep (science.org) 1

sciencehabit writes: On a stormy night in 1816, Mary Shelley had a terrifying dream about a corpse coming to life—a nightmare that inspired her to write Frankenstein. More than a century later, a melody in a dream led Paul McCartney to compose one of The Beatles’s most beloved songs, Yesterday.

Is there something about dreaming that enhances our creativity? Or is it just sleep itself? Scientists say they’re closer to an answer, thanks to an unusual study that used an electronic glove to guide people’s dreams while they slumbered.

To conduct the work, researchers invited 50 volunteers, mostly students and professors, to either stay awake or take a nap in a laboratory at the Massachusetts Institute of Technology (MIT). Those in the nap group laid down with an eye mask, while wearing a Dormio, a glovelike device with sensors that measure heart rate and muscle tone changes to track sleep stages. A computer linked to the device relayed audio cues to inspire the wearers to dream about specific subjects—a process called “targeted dream incubation.”

Overall, volunteers who dreamt about trees scored 78% higher on the creativity metrics than those who stayed awake just observing their thoughts and 63% higher than those who stayed awake thinking about trees. Participants who napped without hearing the prompt still got a creativity boost, but those who dreamed about trees still performed 48% better than them.

The researchers also noticed that the volunteers used the content of their dreams to answer the tests. The person who dreamed that their limbs were made of old wood wrote a story about an oak king with a wood body, for example. The person who dreamed of becoming bigger than trees, meanwhile, listed “toothpick for a giant” as an alternative use for a tree.

Submission + - Google Releases 2023 Workforce Diversity Report Based on Pre-Layoff Numbers

theodp writes: If anyone thought Google's 2023 Diversity & Inclusion Report would shed light on what Google's U.S. workforce numbers look like after the company lays off 12,000 employees (6% of its workforce) who CEO Sundar Pichai explained were "hired for a different economic reality," think again. Readers of Google's recently-released 2023 D&I report are instead provided with fuzzy company-spun percentages (e.g., "We’ve achieved our racial equity commitment goal of increasing leadership representation of Black, Latino and Native American Googlers by 30% – three years ahead of our 2025 aim") that are based on pre-layoff numbers. "Representation data reflects Googlers employed on January 1, 2023," explains Google on page 53 of the report.

In releasing the report, Google did not elaborate on what it made of its restated 2014-2023 tech workforce diversity percentages or its ultimate goals for the larger technical workforce and when they might be achieved. To be fair to Google, while it may have been a pioneer, the practice of spinning company-chosen percentage-based metrics in D&I reports rather than presenting raw workforce numbers — a move straight out of the "How to Lie With Statistics" playbook — has also been embraced by Google's tech giant peers. So, will Microsoft, Amazon, and Facebook — all of whom have announced big layoffs — follow Google's lead and release 2023 workforce reports that fail to reflect their layoffs?

Submission + - SPAM: TikTok Feeds Teens a Diet of Darkness 2

schwit1 writes: Calls to ban TikTok in the U.S. are growing louder. Government leaders are trying to keep the popular China-owned social video platform away from schools public workers, even entire states, on the grounds that users’ data could wind up in the wrong hands.

Data privacy, though, might be less worrisome than the power of TikTok’s algorithm. Especially if you’re a parent.

A recent study found that when researchers created accounts belonging to fictitious 13-year-olds, they were quickly inundated with videos about eating disorders, body image, self-harm and suicide.

If that sounds familiar, a Wall Street Journal investigation in 2021 found that TikTok steers viewers to dangerous content. TikTok has since strengthened parental controls and promised a more even-keeled algorithm, but the new study suggests the app experience for young teens has changed little.

What teens see on social media can negatively affect them psychologically. Plenty of research backs this up. The simplest evidence may be found in my earlier column about teens who developed physical tics after watching repeated TikTok videos of people exhibiting Tourette Syndrome-like behavior.

The Center for Countering Digital Hate, a nonprofit that works to stop the spread of online hate and disinformation, tested what teens see on TikTok.

Last August, researchers set up eight TikTok accounts to look like they belonged to 13-year-olds in the U.S., the U.K., Canada and Australia. For 30 minutes, researchers behind the accounts paused briefly on any videos the platform’s For You page showed them about body image and mental health, and tapped the heart to like them.TikTok almost immediately recommended videos about suicide and eating disorders, the researchers said. Videos about body image and mental health popped up on the accounts’ For You pages every 39 seconds, they added.

After the researchers published their findings, many of the videos they flagged disappeared from TikTok. Many of the accounts that posted the material remain. Those accounts include other videos that promote restrictive diets and discuss self-harm and suicide.

Link to Original Source

Comment Re:Starlink Rural Canada offer (Score 1) 45

Do you use a directional antenna to connect to the tower - those improve the signal a lot, especially as you have a clear line-of-sight?
And use Cellmapper.net service to make sure there aren't any other cell towers nearby (maybe with the beams better directed at your house).

I have a 10' (3m) parabolic reflector behind the transponder. I get a measured gain of 21 to 28 db depending on the band.

I know where every tower is for 100km around me. Cellmapper is ok, not always complete or accurate. Industry Canada maintains a public database of tower information.

Comment Re:Starlink Rural Canada offer (Score 1) 45

Actually speeds from 5G network should go much, much higher than 75Mbps. I guess in your case the main limiting factor is the distance to the tower, since the faster bands use higher frequencies - which consequently have shorter ranges.

But just for comparison, a local operator (Telia) offers here in Finland a nice 5G asymmetrical 1Gbps down 100MBps up connection for 45 euros per month. No data caps, no slowdown.

We get much better data rates in urban areas where there are higher frequency bands available and distances are under 1000 m. My locale is RF transmission challenged with 60' trees all around. I get my signal through a gap where my driveway comes up; at night I can see the tower hazard light on the horizon.

Gigabit service, unlimited data, CAD 70 per month? I wish!

The only reason TELUS is offering the plan I described is to compete with Starlink.

Comment Starlink Rural Canada offer (Score 3, Interesting) 45

Just received an offer from Starlink of CAD 199 (versus $749 normal) for the equipment purchase and $140 per month. Your location has to be approved but it is labelled "rural Canada" which covers a lot of territory. The map shows pretty much everything between latitude 49 N and about 70 N. Tough luck Resolute Bay and Alert.

5G wireless rural internet from TELUS is now unlimited data (no cap, no slowdown) for $95 per month on a two year plan and the equipment is free. Well actually you pay for it over two years but the payment is rebated every month: net zero.

Thank you EM for giving the terrestrial providers some competition.

Mind you, the current $95/month is a "special". Starlink, TELUS, Bell, et al are ready to jack the prices back up as soon as the competitive atmosphere dissipates.

Speeds from the 5G network can be up to 75Mbs but where I am it is typically 30Mbps symmetrical up & down. Also, the 5G is not stand alone (relies on the LTE network for control) and can be unstable at longer distances (I am 12 km from the tower). Apparently, when they upgrade to stand alone 5G, those problems go away.

Comment Bullshit AI (Score 5, Interesting) 55

eh, ChatGPT et al are convincing bullshitters, which is quite impressive if not what AI should be targeting. More accurately ChatGPT is a confabulator because it cannot tell if it is lying or not. It draws on stored information (memories) but is missing a lot and lacks understanding of context so it stitches together a story that may or may not have any congruence with reality.

Confabulation is often associated with diseases such as Alzheimer's, or when there has been memory loss, or as part of the expression of other pervasive mental disorders.

From the Turing Test pov, sure I think it is difficult to tell if I'm talking to a machine or a convincing con man but I wouldn't use this so called AI for anything critical.

The fact that it is commercially useful is food for thought.

Comment Re:5G or satellite - pick one (Score 1) 34

I am 10 km from a tower and my phone sometimes hits a 5G panel.

You are confusing the protocol with the data capacity and range of transmission bands. 5G can be implemented on all of the bands used for LTE but data rates are constrained by frequency of the bands, and as band frequency increases data capacity goes up but object penetration goes down. Practically speaking this means hundreds of metres or less in an urban area for the higher bands. Trees and leaves are effective RF shields above a couple GHz.

With line of sight (no objects in the RF path), there is nothing to block the signal so 5G from a satellite is possible.

There are some other constraints as I believe both LTE and 5G have to self adjust for transmission delays and were not originally conceived to handle satellite round trip times.

Submission + - Ignored by Google-Gallup's K-12 CS Research, Asian Kids Now Rule AP CS A

theodp writes: In 2020, Google released Current Perspectives and Continuing Challenges in Computer Science Education in U.S. K-12 Schools. "This report presents findings from the third study in Gallup and Google’s multiyear, comprehensive research effort to better understand perceptions of computer science and access to computer science learning opportunities in K-12 schools in the U.S," explained Google. "The purpose of these studies is to [1] explore perceptions and understanding of computer science, [2] examine data on access and availability of computer science education for all students in U.S. schools, [3] gauge the extent to which students, parents and guardians, and educators value and prioritize computer science learning and careers." As noted earlier on Slashdot, the Google-Gallup research explored race/ethnicity differences to access and learning opportunities among White, Black, and Hispanic students but oddly chose to ignore Asian students.

Newly-disclosed AP CS exam data for the Java-based AP CS A course — which Amazon is seeking to push into more high school classrooms by spending $15M and enlisting the help of Google and Microsoft — makes Google's decision to exclude Asian students and parents from its research all the more puzzling. A look at the AP CS A exam data with a focus on the numbers and percentages of students who received passing scores reveals that Asian students now top all racial/ethnic groups when it comes to AP CS A success, making their omission from the Google-Gallup study of 'all students' very baffling indeed. In Google's home state of California, 5,423 Asian students received passing AP CS A scores, compared to 1,936 White, 912 Hispanic, and 66 Black students. And in Amazon's and Microsoft's home state of Washington, 870 Asian students received passing AP CS A scores, compared to 624 White, 72 Hispanic, and 14 Black students.

Asian students, who U.S. Dept. of Education stats say make up about 6% of U.S. public high school students, accounted for 41% of all passing AP CSA exam scores in 2022 (up from 29% in 2010) and had a pass rate of 75%. White students, who account for an estimated 45% of HS students, accounted for about 38% of all passing AP CSA scores in 2022 (down from 58% in 2010), with a pass rate of 69%. Hispanic students, who make up 29% of HS students, accounted for 8% of all passing 2022 AP CSA scores in 2022 (up from 4% in 2010) and had a pass rate of 45%. Black students, who make up an estimated 15% of HS students, accounted for only 2.1% of passing AP CSA exam scores in 2022 (up from 1.7% in 2010) and had a pass rate of 31%. Finally, a quick look at the number of AP CS A exams passed by state shows striking gaps in participation and outcomes across groups and surprisingly low participation numbers across all groups in many states despite a 10+ year concerted effort by the nation's tech giants and K-12 CS nonprofits to address the low number and diversity of AP CS A students.

Submission + - Kandinsky 2.1 – an updated neural network for generating images on demand (altervista.org)

FactoidBreaker writes: Sberbank has introduced Kandinsky 2.1 – an updated neural network for generating images on demand. Now it works better with high resolution and complex objects. Sberbank announced a major update to the Kandinsky neural network, which generates images from text descriptions. The neural network can change existing images on demand, combine several images, draw fragments, create images in the same style as suggested by the user, and more. It understands requests in 101 languages

Submission + - Snapchat's AI Chatbot Is Now Free For All Global Users (techcrunch.com)

An anonymous reader writes: Snapchat’s AI chatbot is now opening up to a global audience, the company announced today at its Snap Partner Summit. Initially launched in February, the feature originally allowed Snapchat’s paid subscribers to chat with an AI chatbot powered by OpenAI’s GPT technology directly in its app. Now it will be available for free. To date, users have sent nearly 2 million messages per day using the chatbot, Snap noted. With today’s global expansion, the feature is also being upgraded with new functionality, including the ability to add My AI to group chats, get recommendations for places on Snap Map and Lenses, and share Snaps with My AI and receive chat replies.

Later, My AI will be able to respond with unique “generative” Snaps back, instead of just chat replies, the company also said, to keep the visual conversation going. The idea to integrate AI into the Snapchat app was originally intended to give users another way to engage in the app while taking advantage of the growing consumer demand for ChatGPT-like experiences. The company suggested the feature could be used to do things like suggest birthday gift ideas for a BFF, plan a hiking trip, suggest dinner recipes or write a poem for a friend, among other things.

Comment Re:What about users? (Score 3, Interesting) 81

Surely they must keep track of who has an account somehow. I see that they do take money from their customers, since they sell access.

Mullvad can do that without being able to provide police with any useful identifying info.

Undoubtedly the police are interested in correlating sessions to user ids to real people.

Say Mullvad is both truthful and competent in their claim of not logging user data then ...

- The company as a minimum will record payments to user ids but that does nothing to identify which real person did what online. Mullvad does not have to keep any payment provider info after the payment clears so connecting a userid to a real person via payments may not be possible. If the subscriber has used an anonymous payment method and has used a throw away email address then even if payment records exist, they may not be useful to id a real person.

- When a user turns up a VPN session, it's userid is validated against a subscription record and the session is allowed to be created without further need of the userid. There is no technical or billing requirement to log session-userid information.

- The active session of necessity knows the user's source ip but this will only persist for the duration of the connection and of course there is no need to log the ip-session info. The real time session connection information will be available to a sysadmin in some form, although I suppose this could be obfuscated. This would not be useful for historical sessions.

To find the real session userid, the police would have to either have a warrant to force the provider to give them access to active VPN information while the target is online or they would have to enlist NSA level sniffing of the provider's internet connections for command and control, and the vpn node.

Again, even if they get a userid for a session that may not be sufficient to connect to a real person.

Slashdot Top Deals

He has not acquired a fortune; the fortune has acquired him. -- Bion

Working...