Hashtag Trending

Jim Love
  • 5,0 (12)
  • Технологии
  • Ежедневно

A daily news program covering the top stories in technology with a weekend in depth interview.

  1. -13 ч

    GPT-6 Astra Launches Amid ChatGPT Outage | NVIDIA Buys Hugging Face | Meta Cuts AI Costs

    OpenAI calls GPT-6 Astra a generational leap, with company president Greg Brockman declaring, "Welcome to the AGI era." But OpenAI chief scientist Jakub Pachocki is sounding a very different note. In his essay "An Alien Mind," Pachocki warns that no one is prepared for the consequences of rapidly rising machine intelligence. Jim Love examines Astra's critical cybersecurity capabilities, reduced chain-of-thought monitorability, potential to evade safety monitors during adversarial testing, AI-assisted training and staged rollout to paying ChatGPT subscribers. He also considers how AI helping to train its successors echoes a warning in Eliezer Yudkowsky and Nate Soares's bestseller, "If Anyone Builds It, Everyone Dies." Next, a Gamers Nexus investigation alleges that LG smart TVs catalog devices on their networks, monitor screen content and transcribe or retain conversations. The findings raise serious questions about advertising data, cybersecurity and whether television data could become evidence in a lawsuit. LG previously said its televisions do not record ambient conversations. Finally, physicist and YouTuber Sabine Hossenfelder says organizations on both sides of the AI debate are paying influencers to promote their messages—sometimes supplying the wording and sources themselves. The episode closes with a clear statement about Hashtag Trending's editorial independence. 00:00 Headlines and introduction 00:50 GPT-6 Astra and the AGI debate 02:01 Jakub Pachocki's "An Alien Mind" warning 03:26 Hidden reasoning and weaker monitoring 05:30 Safety slowdowns and Astra rollout 06:15 AI training its successors 07:38 LG TV data-collection investigation 09:42 Advertising value and security risks 11:22 Paid influencers in the AI debate 13:23 Editorial independence and closing

  2. -3 дн.

    Project Synapse: Astra, AGI Hype, Model Costs, and AI's Real-World Utility (Plus Education & Safety)

    Jim Love hosts Project Synapse with John Pinard and University of Waterloo professor Doug Sparks, reviewing a week of intense AI model releases led by OpenAI's Astra (called a "generational leap" by Greg Brockman) and debating what "AGI-ish" could mean, including Sam Altman's definition tied to commercially performing a wide range of human tasks and claims about creating new knowledge. They discuss how to test creativity, using AI for scenario planning and backcasting, and frustrations with inconsistent behavior and feature gaps across ChatGPT web/desktop/voice experiences. The conversation covers rising costs and slow runtimes for complex tasks, while noting cheaper, strong-performing alternatives from Google and Meta. They also address AI in education, arguing for AI literacy and assignments emphasizing critical evaluation, and raise governance and safety concerns about opaque government oversight, chain-of-thought removal, models training models, and hacking risks. 00:00 Welcome and Guests 01:32 Labor Day Model Frenzy 02:39 Astra and AGI Claims 05:49 Defining and Testing AGI 09:01 Scenario Planning with AI 13:56 Utility Over Creativity 17:27 Meta AI Rollout Lessons 20:52 Cost and Speed Reality Check 22:06 Voice Mode Limitations 27:04 Anthropic vs Google vs Meta 30:22 Cheap Models and Real Use 31:57 Planning With AI Models 33:20 AI As Learning Coach 34:30 Cheating Or New Tool 35:36 Teaching Critical Evaluation 38:03 Education Factory Debate 41:18 NYC AI Policy Experiment 43:13 Model Safety And Oversight 47:46 Transparency Versus Politics 51:12 Global Governance Dilemma 52:59 Risks Data Centers Backlash 56:20 Dark Futures Scenario Planning 57:21 Next Year Projects Wrap

  3. -4 дн.

    GPT-6 Astra Launches Amid ChatGPT Outage | NVIDIA Buys Hugging Face | Meta Cuts AI Costs

    OpenAI unveiled GPT-6 Astra on the same day ChatGPT and Codex suffered a widespread outage lasting just over two hours. There is no evidence the outage was caused by Astra, but the timing was striking. OpenAI is describing Astra as a major step forward, with stronger agentic abilities, advanced cybersecurity capabilities, and what Sam Altman has called the potential to "actually invent new things in a way that matters," something he described as "a very AGI-like thing." Earlier OpenAI models also played a large role in supervising Astra's training, another indication of how AI is increasingly being used to develop the next generation of AI systems. NVIDIA is buying Hugging Face for $12.9 billion, strengthening its position in the open-source AI ecosystem just as some of its largest customers are developing their own chips. Jensen Huang says Hugging Face will remain open to competing models, clouds and hardware. Meta is attacking another part of the market with Muse Spark 1.3, offering frontier-level performance at dramatically lower prices than many major Western models. Its Contributor tier is even cheaper, although developers must allow Meta to use their prompts and responses for training. And LibreOffice 26.8 passed one million downloads in its first week, with one unusual selling point getting attention: no built-in AI. The Document Foundation summed up the mood in a post titled, "Yes, no AI is now a feature." Hashtag Trending for Friday, September 4, 2026, hosted by Jim Love. 00:00 Headlines and Intro 00:35 GPT-6 Astra Launch and ChatGPT Outage 01:36 Astra, AGI Claims and AI-Assisted Training 02:51 Astra Access and Rollout 03:24 NVIDIA Buys Hugging Face for $12.9 Billion 05:14 Meta Cuts the Cost of Frontier AI 09:32 Muse Spark Benchmarks and Strategy 11:44 LibreOffice Bets on No Built-In AI 13:23 Wrap Up and Sign Off

  4. -5 дн.

    Google Dodges Antitrust Breakup, Florida Pulls Flock Cameras, Meta Tests Data Center Robots

    im Love covers the biggest tech stories for Thursday, September 3. Google has again avoided a court-ordered breakup. A federal judge rejected the Justice Department's attempt to force the sale of Google's AdX advertising exchange, choosing behavioural remedies instead. The two sides now have 30 days to work on final judgment language, while the DOJ can still appeal. Protect Democracy is suing for access to the rules governing secret federal reviews of frontier AI models. The case raises a larger question about how much influence any administration should have over the AI systems increasingly used to summarize news and information. Florida is ordering Flock and other automated license-plate readers off state highway property, a largely symbolic move that sends a much bigger message about AI-powered police surveillance. Texas is also stopping new state funding for Flock cameras. Meta is testing robots inside data centres that could eventually automate as much as 80 percent of some technician workloads. That comes after Meta's earlier attempt to dramatically automate software and development work with AI ran into problems. And New York City is imposing a one-year moratorium on student-facing generative AI from pre-K through Grade 8. But the more important part may be what happens next: older students will be taught AI literacy with an emphasis on critical thinking, bias, evidence and questioning AI-generated answers. 00:00 Headlines and Intro 00:39 Google Antitrust Remedies 03:08 FOIA Fight Over Secret AI Reviews 05:54 Florida Pulls Flock Cameras 07:52 Meta Tests Data Center Robots 10:21 NYC Limits Classroom AI 12:26 Wrap Up and Sign Off

  5. -6 дн.

    Microsoft 365 Outage Spreads as AI Bug Reports Overwhelm Linux

    A Microsoft 365 outage that began with Exchange Online spread across Teams, OneDrive, SharePoint, Copilot, Purview, Defender XDR and administrative tools. Microsoft traced the disruption to an authentication configuration error affecting shared infrastructure. Linux kernel maintainers are being overwhelmed by automated and AI-generated bug reports, many of them duplicates or poorly validated. The problem has become serious enough that maintainers are using frontier AI models to help triage the reports and considering whether rarely used legacy code should simply be removed. An AV-Comparatives test using 250 live phishing sites found major browsers blocking approximately 75 to 79 per cent. Several dedicated security products blocked between 94 and 96 per cent, demonstrating why browser protection alone may not be enough. Amazon is also pouring money into AI and cloud infrastructure while reducing corporate employment. At the same time, US Government Accountability Office data shows a growing number of Amazon employees receiving SNAP food assistance during 2025. Jim Love covers these stories on Hashtag Trending for Wednesday, September 2, 2026. 00:00 Today's tech headlines 00:26 Microsoft 365 outage spreads 01:41 How shared authentication widened the outage 02:07 AI bug reports overwhelm Linux maintainers 03:40 Linux uses AI to triage AI reports 04:22 Browser phishing protection tested 05:30 What IT teams should do next 06:27 Amazon's AI spending and labour divide 08:40 Wrap-up and where to reach us

См. все (923)

5
из 5
Оценок: 12

Об этом подкасте

A daily news program covering the top stories in technology with a weekend in depth interview.

Информация

  • Автор
    Jim Love
  • Годы выхода
    2017 - 2026
  • Выпуски
    923
  • Ограничения
    Без ненормативной лексики
  • Сайт подкаста
    Hashtag Trending

Вам может также понравиться