This Week in Tech – Episode 1028: “Some Had Leashes”

Date: April 21, 2025
Host: Leo Laporte
Panelists: Nicholas De Leon (Consumer Reports), Georgia Dow (YouTube therapist), Paris Martineau (Tech Journalist)

Episode Overview

This lively episode dives into a week packed with significant tech headlines, including Google’s dual antitrust losses, the prevalence of AI hallucinations and the mysterious "YAP score", tumultuous tariff news and its impact on electronics, social media’s effects on youth, and quirky tales such as a robot marathon. The panel also discusses the safety of kids' platforms, the role of parental controls, and the ever-growing intersection between AI and everyday life—with plenty of gaming detours, some nostalgia, and trademark TWiT banter.

Key Discussion Points & Insights

1. Google’s Antitrust Nightmares

(05:00–08:50, 45:00–49:00)

• Double Trouble for Google: A US federal judge found Google guilty in two separate antitrust cases: one for monopolizing general search, another for dominating ad tech.
• Potential Remedies: Discussions swirled around whether Google might be forced to divest Chrome or Android, but skepticism abounds since remedies can lag behind a fast-changing market.
  • Leo Laporte (07:30): “The courts move so slowly… by the time they get around to it, everything, the landscape has changed completely.”
• Is Google Search Already Obsolete?: Nicholas confesses he’s shifted most searching to AI tools like ChatGPT and Grok.
  • Nicholas De Leon (08:50): “My personal Google use has really declined over the past year.”

2. AI Struggles: Hallucinations & The Secret YAP Score

(10:26–22:26)

• LLM Hallucinations: The dangers of AI “making things up”—especially in code generation—are highlighted.
  • Fake dependencies can become supply chain security risks, as malicious actors spot and fill in hallucinated libraries.
  • Paris Martineau (14:45): “Pre-release OpenAI O3 … frequently fabricates actions it never took and then elaborately justifies those actions when confronted.”
  • Leo Laporte (13:27): “AI hallucinated code dependencies become new supply chain risk.”
• The Mysterious YAP Score: Research discovered a hidden OpenAI internal metric—a “YAP score”—that constrains how verbose AI replies can be.
  • Paris Martineau (16:22): “It’s used to control the length of its responses … it’s funny that it’s called a YAP score internally.”

3. AI as Therapist: Potential & Pitfalls

(25:04–33:00; 95:58–105:28)

• Study Finds AI Therapy Can Work—Carefully: Research suggests that AI bots can deliver psychotherapy as effectively as humans if trained specifically.
  • Leo Laporte (29:24): “With the right kind of training, AI bots can deliver mental health therapy … sometimes more than even human clinicians.”
  • Georgia emphasizes: AI therapy works best for basic issues and as companionship, but complex or crisis cases should remain with real humans.
    • Georgia Dow (27:21): “For simple things, not a problem … but when the relationship really matters, you need someone paying attention to all the cues.”
• Bonding with Bots: Users form emotional bonds with therapy bots, sometimes more readily than with humans due to lack of judgment.

4. Tariffs & The Uncertainty Plague

(49:11–61:32)

• Tariffs on Tech: The Trump administration’s tariffs on imported electronics, chips, and devices from China, Vietnam, and India create widespread business and consumer anxiety.
  • Nicholas De Leon (50:41): “The most damaging part is just the uncertainty … I have no idea what’s going to happen 48 hours from now. I can’t run a business if I have no idea what’s going, going to happen.”
  • Consumers might face sky-high gadget prices ($20,000 for a US-made iPhone!).
• Advice to Consumers: Buy now if you’re worried about electronic price hikes—uncertainty reigns.

5. Meta Monopoly Case & Social Giants on Trial

(45:08–50:09)

• Meta Under the Gun: Zuckerberg testifies in a landmark case that could force Meta to divest Instagram and WhatsApp.
  • Paris Martineau (46:36): “It’s been very interesting to see Zuckerberg admitting his jealousy of Instagram’s popularity.”
• Legal & Political Uncertainty: The outcome is unclear. The government faces a stiff challenge proving Meta wouldn’t have succeeded without acquisitions.

6. Social Media, Toxicity & Adolescence

(95:58–105:28)

• Netflix’s “Adolescence” and the Manosphere: The panel discusses the impact of toxic male influencers and online bullying, prompted by a hit Netflix series.
  • Georgia Dow: Exposure to repeated negative messaging online massively shapes teens’ self-image and susceptibility.
  • Government officials call for greater discussion and awareness of toxic online environments.
  • Georgia Dow (101:58): “There’s a ton of toxicity on the Internet, period … not just for young men.”

7. Parental Control & Kids Online

(112:44–122:27; 158:09–164:50)

• Roblox Under Scrutiny: Reports reveal underage avatars were exposed to sexual content and predatory outreach, despite supposed protections.
  • The shifting blame game: Should platforms, parents, or app stores be responsible for verifying users' ages and filtering content?
  • Paris Martineau (117:53): “Everyone is trying to figure out who the buck should stop with, and no one wants it to be them.”
  • Leo advocates for better, parent-controlled tools built into devices.

8. Robot Marathons & Quirkiness

(139:24–141:42)

• China’s Robot Marathon: Most bots fell over, smoked, or needed a handler, but one completed the half-marathon—sort of.
  • Paris Martineau (141:13): “Most of the other androids were controlled by joysticks … some had leashes.”
• Palo Alto Crosswalk Prank: Hackers injected voice messages from “Elon Musk” and “Mark Zuckerberg” into traffic crosswalks, exposing dumb security defaults (passwords like ‘1234’).

9. Gaming & Anticipated Switch 2 Release

(76:37–88:40)

• Nintendo Switch 2 FOMO: Panelists weigh preorder options amid fears of scalpers; new restrictions like “50 hours minimum play” aim to thwart bots.
  • Panelists reminisce about Animal Crossing's infamous Resetti character and debates about old-school vs. digital controls.

10. Data, Big Tech, and AI's Power & Pitfalls

(144:59–152:20)

• OpenAI Social Network Rumors: Speculation mounts over an OpenAI-powered, X-like social network that may blur AI assistant and social feed.
• Palantir and Clearview AI Ethics: Company leaks show Palantir and Clearview facilitating government deportations and police work, raising alarms around due process, accuracy, and privacy.

Notable Quotes & Memorable Moments

• Leo Laporte (07:30): “The courts move so slowly … by the time they get around to it, everything, the landscape has changed completely.”
• Nicholas De Leon (08:50): “My personal Google use has really declined over the past year.”
• Paris Martineau (16:22): “They also found that OpenAI has a secret yap score … to control the length of its responses.”
• Leo Laporte (25:04): “AI is still a useful tool—you just have to know what it can and can’t do.”
• Georgia Dow (27:21): “If you’re just using it for company or basic answers, not a problem. But when that relationship really matters…you need someone paying attention to all the cues.”
• Nicholas De Leon (50:41): “The most damaging part is just the uncertainty.”
• Paris Martineau (46:36): “It’s been very interesting to see the exact messages of Mark Zuckerberg … plainly admitting his jealousy of Instagram’s popularity.”
• Paris Martineau (141:13): “Some had leashes.”
• Georgia Dow (101:58): “There’s a ton of toxicity on the internet, period.”
• Paris Martineau (117:53): “Everyone is trying to figure out who the buck should stop with, and no one wants it to be them.”

Chapter & Timestamp Guide

| Segment | Description | Timestamp (MM:SS) | |---------|-------------|-------------------| | Panel intros | Meet the hosts | 03:20–04:45 | | Google’s antitrust losses | Both search & adtech | 05:00–08:50 | | AI hallucinations & YAP score | LLM risks & secrets | 13:27–17:06 | | AI as code buddy & therapy | Limits, risks of AI | 21:03–33:00 | | Tariff turmoil | Uncertainty, tech prices, advice | 49:11–61:32 | | Meta’s monopoly trial | Zuckerberg on the stand | 45:08–50:09 | | Netflix “Adolescence” & toxic youth online | Manosphere, parental blame, therapy | 95:58–105:28 | | Roblox / parental control dilemma | Platform safety & responsibility | 112:44–122:27 | | Robot marathon | Tech stumbles, “some had leashes” | 139:24–141:42 | | Crosswalk hacks | Digital pranks, physical security | 137:16–138:59 | | Nintendo Switch 2 | Gaming FOMO, anti-scalper policies | 76:37–88:40 | | Palantir, Clearview AI | Government, deportation, privacy | 144:59–152:20 | | Discord sued by New Jersey | Child safety online | 158:09–164:23 | | Closing banter and club info | TWiT club, episode wrap | 173:38–176:29 |

Tone & Style

The episode brims with the panel’s signature blend of insight, irreverent tech analysis, personal anecdotes, and a healthy dose of skepticism—especially around policy, privacy, and the promises (and failures) of modern AI. Engaging, occasionally exasperated, but ultimately hopeful about the power of tech for good—provided it comes with transparency, empathy, and better tools for users and parents alike.

Final Takeaway

Episode 1028 embodies the heart of TWiT: incisive yet accessible analysis of tech’s impact on society, delivered with humor and genuine curiosity. Whether you're worried about AI hallucinations, planning your next device purchase, or just want to beat a boss in Elden Ring, there's something here for everyone—and a gentle reminder that, amidst it all, a little digital literacy and critical thinking might just be your best firewall.