OpenAI hits back at Elon Musk, 90 percent of statistics published on the Internet aren’t true, 30,000 Fidelity customers in the US have their data stolen and those hospital alarms sounds may not save lives – and they are responsible for a lot of deaths.
All this and more on the “lies, damned liars and internet statistics” edition of Hashtag Trending. I’m your host, Jim Love, CIO of IT World Canada and TechNewsDay in the US.
So, it turns out that Elon Musk’s lawsuit might have left out a few details. A big part of Musk’s lawsuit accuses OpenAI of changing from a not for profit to a for profit company and of giving Microsoft control of AI.
But documents released by OpenAI claim that the board didn’t want to try to become a for profit company, because a not for profit could not raise the amount of money that would be required to pursue their mission of Automated General Intelligence. They also maintain that Musk knew that all along an didn’t object.
Not only did he not object, but again, according to the documents released by OpenAI, Musk actively wanted to merge the company with Tesla, a company he controls.
The board was concerned about turning over control to any one person, especially if that person is Elon Musk.
As to Musk’s claim that the board has given away the rights to Artificial General Intelligence and made Microsoft a “defacto owner” of AGI, OpenAI has already pointed out in a staff memo yesterday, that Microsoft’s agreement does not entitle that company to any ownership or right to AGI. Further, that memo indicated that it was in the board’s sole discretion to say what was AGI and what was not.
It does have access to the “standard” generative AI products produced by OpenAI, like GPT 4. But without the 10 billion dollars brought in from Microsoft, everyone, including Musk, knew that OpenAI would not have been able to have the computing power to pursue their AI research and development.
Or as ChatGPT would have said if we asked it to summarize Musk’s position as if they were William Shakespeare, “methinks the Elon doth protest too much.”
Source: AIGrid (YouTube) and the Verge
A recent set of takeaways from a Europol report has received widespread quotation across the internet, including a statement that 90% of Online content will be AI-Generated by 2026.
The only problem is, that quote is nowhere to be found in the report, according to an article published by a company called Oodaloop.
The article also contains some very interesting stats from a report by the UK based firm Public First.
62% of respondents supported the creation of a new government regulatory agency, similar to a medical regulatory agency, to regulate the use of new AI models.
Overall, 32% thought advanced AI would make us safer, compared to 18% who thought it would make us less safe. When asked about specific risks from advanced AI, the most important were perceived to be increasing unemployment (49%)
Even though this is a UK report, it has some interesting research in it. We’ll post a link to those in the show notes.
But as someone who puts together a daily newscast, there was a real lesson here. Just because a lot of people say a statistic exists, doesn’t mean that it does.
Sources include: Oodaloop
Nearly 30,000 Fidelity Investments Life Insurance customers’ personal and financial information is feared stolen in a suspected ransomware attack, following a breach into Infosys’ IT systems last fall. The compromised data likely includes bank account and routing numbers, credit card numbers and security codes, names, Social Security numbers, states of residence, and dates of birth. This breach could potentially enable identity theft scams or unauthorized financial transactions.
Infosys, an Indian tech services giant, experienced the cybersecurity incident affecting its US subsidiary, Infosys McCamish Systems (IMS), which led to the shutdown of some applications and IT systems. LockBit, a notorious ransomware gang, claimed responsibility for the intrusion. This incident marks another significant data breach involving Infosys, following a similar disclosure last month related to a Bank of America data leak.
The cyberattack occurred between October 20 and November 2, disrupting services provided by Infosys to both Fidelity and Bank of America. Fidelity has been working with IMS to investigate and contain the event, implement remedial measures, and safely restore services.
Sources include: The Register
Hospital workers are exposed to up to 1,000 alarm noises per shift, leading to a phenomenon known as “alarm fatigue.” This sensory overload has been linked to hundreds of deaths annually. A new study suggests that replacing standard monotonous hospital alarms with more musical ones could significantly reduce alarm fatigue while making key equipment less annoying.
The study, which involved experimenting with different musical sounds for hospital alarms, found that only 15% of all alarms in the critical care unit environment were clinically relevant. The high number of false alarms contributes to alarm fatigue, a desensitization caused by sensory overload, potentially leading to missed alarms. Between 2005 and 2010, the US FDA reported 566 alarm-related deaths.
Researchers Joseph Schlesinger, an anesthesiologist at Vanderbilt University Medical Center, and Michael Schutz, a music cognition researcher at McMaster University, have been exploring how timbres might allow softer sounds to command the attention of busy medical personnel. They discovered that sounds with a “percussive” timbre, containing short bursts of high-frequency energy, stand out even at low volumes. In contrast, loud, “flat” tones without high-frequency components tend to get lost.
In their study, 42 participants were presented with six alarms: half designed according to a standard alarm and half with a new timbre based on the sound of a xylophone. The researchers assessed participants’ perceived annoyance with the different alarms and their ability to recognize them. The results showed that complex percussive timbres were considered less annoying in 88% of instances compared to standard tones used in medical devices. Alarm melodies in an acoustically rich timbre were no more difficult to identify than standard hospital beeps.
This research indicates that musical timbres can significantly reduce perceived annoyance without harming alarm learnability. It represents a step towards improving alarm design while addressing the issue of excessive alarm sounds among medical devices. Future research will explore how different timbres affect other important perceptual issues, such as alarm detectability.
Sources include: NewAtlas.com
The study was published in the journal Perioperative Care and Operating Room Management.
That’s our show for today.
Love your comments.
Send us a note at jlove@itwc.ca or drop us a comment under the show notes at itworldcanada.com/podcasts – look for Hashtag Trending.
Thanks for listening and have a Thrilling Thursday.