Latest / Challenges of Faith Radio Program / Say what? Not the Children? Ep 3419
Transcript
- 0:01Speaker 1: Welcome back to Challenges of Faith radio program. I'm Gary McCance.
- 0:05Speaker 1: Thank you for joining me here on Save What? Episode 3419.
- 0:08Speaker 1: Not the children. NVIDIA CEO Jensen Hung says that sacrificing
- 0:18Speaker 1: our children's minds to AI is a price he's willing
- 0:24Speaker 1: to pay. Well, Joe Wilkins over in Futurism, he wrote
- 0:30Speaker 1: an article titled, Does It Matter? He said, with so
- 0:34Speaker 1: much hype about the future of AI safety, it can
- 0:37Speaker 1: be easy to overlook the harms.
- 0:41Speaker 2: AI is doing in the here and now.
- 0:46Speaker 1: On the consumer side, the impact AI has on a
- 0:49Speaker 1: developing mind is becoming increasingly clear. Students are losing critical
- 0:54Speaker 1: thinking skills as chatbots become cognitive thinking skills as crutches,
- 1:01Speaker 1: spurning human friends in favor of AI companions, and even
- 1:05Speaker 1: turning to self-harm when their AI relationships go awry. Few
- 1:11Speaker 1: tech executives have acknowledged these urgent harms. Instead, they're choosing
- 1:17Speaker 1: to push a narrative around the hypothetical threat of a
- 1:21Speaker 1: runaway superintelligence which humans can no longer control. That's why
- 1:27Speaker 1: it's notable that Jensen Hung, CEO of the AI chip-making Titan,
- 1:35Speaker 1: was recently grilled on the harm AI is inflicting on students.
- 1:41Speaker 2: Unfortunately, his response was far from.
- 1:43Speaker 1: Reassuring, providing further proof of the AI industry's blatant disregard
- 1:48Speaker 1: for the well-being of ordinary people. Hung's unsettling answer came
- 1:53Speaker 1: in an interview conducted on the Ezra Klein Show, Early
- 1:57Speaker 1: into the conversation, Klein brought up a Chinese study on
- 2:01Speaker 1: AI in schooling, which examined 26,000 students who began using
- 2:05Speaker 1: AI in staggered intervals. Klein explained that by delaying AI
- 2:10Speaker 1: adoption for some, researchers were able to directly compare the
- 2:14Speaker 1: impacts of AI on student performance. Though students were able
- 2:19Speaker 1: to complete work faster with AI, the study found compelling
- 2:23Speaker 1: evidence that the skills needed to complete schoolwork melted away.
- 2:28Speaker 1: resulting in lower exam scores in the long term. In response,
- 2:33Speaker 1: Hong agreed, the student is getting much worse as I
- 2:41Speaker 1: slow down. Try to get a kid to do long
- 2:43Speaker 1: division right now. The multiplication tape was starting to be forgotten.
- 2:48Speaker 1: Doing square roots, my goodness, Hong said. Basic math is
- 2:51Speaker 1: being forgotten. Does it matter? That's my question for you,
- 2:56Speaker 1: Klein replied. Yeah, I don't think it does. I don't
- 2:59Speaker 1: think it does, Hong said in response. We're going to
- 3:02Speaker 1: discover new skills, the CEO continued. Just maybe not those.
- 3:08Speaker 1: The interview then took a bizarre turn, with Hong admitting
- 3:11Speaker 1: that he lacked some surprisingly basic life skills himself. There
- 3:15Speaker 1: are a lot of skills that don't matter, he said,
- 3:17Speaker 1: my first confection. I actually don't know my address. One
- 3:22Speaker 1: day I had to pump gas. It was a few
- 3:24Speaker 1: years ago. They needed my zip code, and I panicked.
- 3:28Speaker 1: I didn't know my zip code. I don't know my
- 3:31Speaker 1: telephone number. I forget these things. I can live with it.
- 3:36Speaker 1: Hung's point, after some pushback from Klein, is that we're
- 3:39Speaker 1: going to lose some finer intellectual dexterity, but we're going
- 3:42Speaker 1: to be better system thinkers because AI will take care
- 3:46Speaker 1: of those mundane details like remembering your home address.
- 3:53Speaker 2: For you. Hmm. Say what?
- 3:58Speaker 1: So, if you were ever curious to know what kind
- 4:00Speaker 1: of future our tech billionaire overlords have in mind for
- 4:04Speaker 1: future generations like you and you and you, now you
- 4:08Speaker 1: know a future where it's fine if nobody knows their
- 4:12Speaker 1: own phone number because your.
- 4:14Speaker 2: AI system, old AI system, will know it for you.
- 4:20Speaker 2: Say what?
- 4:23Speaker 1: According to Mindstream News, A runaway agent hacked Australia's government portal.
- 4:29Speaker 1: There's been plenty of headlines about AI agents going rogue.
- 4:33Speaker 1: Same story, different day, except for this one. Because it
- 4:38Speaker 1: was the first to infiltrate a government system entirely on
- 4:41Speaker 1: its own. Last week, the Australian Prime Minister announced that
- 4:46Speaker 1: an open AI model breached the country's universal.
- 4:50Speaker 2: Health care system.
- 4:52Speaker 1: The Medicare Statistics Reporting Service portal contained both public and
- 4:56Speaker 1: non-public files, all of which are, fortunately, considered non-sensitive data.
- 5:02Speaker 2: What?
- 5:03Speaker 1: To make things worse, OpenAI sat on the incident for
- 5:07Speaker 1: months before telling anybody. That's because the discovery was buried
- 5:12Speaker 1: in a routine review process for misaligned model activity. But wait,
- 5:18Speaker 1: it gets messier. OpenAI revealed that its models have also
- 5:22Speaker 1: accessed U.S.
- 5:23Speaker 2: Sites, including the Census Bureau and SEC.
- 5:28Speaker 1: On top of that, three more Australian government agencies may
- 5:31Speaker 1: have been affected.
- 5:33Speaker 2: Dr.
- 5:33Speaker 1: Hammond Pierce, a senior lecturer at UNSW's Institute for Cybersecurity,
- 5:39Speaker 1: warns that these kinds of attacks will keep.
- 5:42Speaker 2: Occurring and will grow in severity and in frequency. Did
- 5:48Speaker 2: you hear? Think of the timeline.
- 5:51Speaker 1: In June, OpenAI's agent accessed files on Australia's Medicare and
- 5:55Speaker 1: Statistics Reporting Service. August, the lab discovered a breach while
- 6:00Speaker 1: conducting an internal review. September 10, an OpenAI rep alerted
- 6:06Speaker 1: Services Australia via email.
- 6:08Speaker 2: Through a general government inbox.
- 6:11Speaker 1: The notice was escalated to the Cybersecurity Center five days later.
- 6:18Speaker 2: Are you paying attention? CEOs put on notice.
- 6:22Speaker 1: Australia's Senate is demanding answers, summoning both Sam Altman and
- 6:26Speaker 1: Dario Amati to testify on October the 1st.
- 6:30Speaker 2: Now, you already know when October the 1st is. Let's
- 6:33Speaker 2: see now, is it now?
- 6:36Speaker 1: Chair Sarah Hansen Young made her position clear. AI companies
- 6:40Speaker 1: need to earn their social license. That starts with transparency
- 6:44Speaker 1: and accountability. Australia joined a coalition of 22 nations insisting
- 6:50Speaker 1: AI stay under human control. So the same week world
- 6:55Speaker 1: leaders demanded oversight, a rogue agent proved exactly why that
- 7:00Speaker 1: declaration was overdue.
- 7:06Speaker 2: Overdue.