WEBVTT

1
00:00:10.000 --> 00:00:24.784
Welcome to the Global Advisors A.I Voices podcast. I'm Maya.
Today is Wednesday, the eighteenth of November 2026.
This is your morning catch-up on A.I. Please like and follow the show for your daily A.I update.

2
00:00:26.833 --> 00:00:49.254
Voter backlash against artificial intelligence infrastructure is gaining momentum in Ohio, threatening major investments like Amazon's data center project in Sunbury. In a decisive local vote, seventy-eight percent of Sunbury residents approved a charter amendment banning large data centers that consume at least twenty-five megawatts of electricity.

3
00:00:49.434 --> 00:01:00.464
This local victory is part of a wider trend, with Ohio voters approving restrictions or bans on data centers in numerous separate ballot issues during the recent election.

4
00:01:00.814 --> 00:01:25.646
The surge in turnout has broader implications for state-level policy. Austin Baurichter of Conserve Ohio predicts that the high number of votes cast will likely push the signature threshold for a potential state constitutional amendment on data centers above four-hundred-fifty-nine-thousand. This higher bar makes it more difficult, though not impossible, for advocates to place a statewide ban on the ballot.

5
00:01:25.996 --> 00:01:44.556
Conserve Ohio is actively pushing for a full moratorium on new data center construction. Baurichter criticizes the incoming governor's proposal for a 'conditional' moratorium, describing it as 'political doublespeak' that fails to address the community concerns driving the local bans.

6
00:01:44.736 --> 00:01:55.424
The group argues that only a complete halt to new projects will satisfy voters who feel their infrastructure and quality of life are being compromised by rapid tech expansion.

7
00:01:57.823 --> 00:02:16.682
In an interview with Silicon Valley Girl, Instagram head Adam Mosseri stated that he looks for three key qualities in hires across all functions: grit, quick learning, and self-awareness. Mosseri noted that the 'quick learner' attribute has become even more important due to the rapid pace of technological change.

8
00:02:17.032 --> 00:02:34.632
Mosseri observed that technology shifts are causing the boundaries between distinct job roles, such as engineering, design, and data science, to blur. He explicitly prefers hiring inexperienced candidates who are improving rapidly over senior candidates who have stopped evolving.

9
00:02:34.982 --> 00:02:49.126
According to Mosseri, A.I has increased both the value and the demand for the willingness to learn new things. This suggests that as tools evolve, the capacity to adapt may outweigh static expertise in hiring decisions.

10
00:02:51.525 --> 00:03:19.548
OpenAI has released GPT-6.2 Astra and Sol, marking a significant shift in pricing and performance for its frontier models. The company cut Astra's standard input and output token prices by twenty percent, while rating Astra High for A.I self-improvement. The system card assesses both models as Critical for cybersecurity and High for biological and chemical capabilities, with only Astra meeting the High threshold for self-improvement.

11
00:03:19.898 --> 00:03:45.520
In terms of performance, OpenAI reports that Astra gained five-point-two points on the Agents' Last Exam version one-point-one compared to its predecessor, GPT-6 Astra. On internal GDPval version two, Astra improves six-point-two percentage points over GPT-6 Astra. These gains suggest a more efficient model that delivers higher quality results with lower computational overhead.

12
00:03:45.870 --> 00:04:05.112
A third-party video review claims Astra scores significantly higher than Anthropic's Fable 5.2 at a fraction of the cost, suggesting potential market disruption. While this is an independent opinion rather than an official benchmark, it highlights the competitive pressure OpenAI is facing.

13
00:04:05.292 --> 00:04:15.639
The combination of lower prices and higher self-improvement ratings positions Astra as a strong contender in the race for efficient, high-performance A.I agents.

14
00:04:18.038 --> 00:04:39.116
In a recent interview with Silicon Valley Girl, Adam Mosseri, head of Instagram, explained how he uses A.I to write emails that sound like him by comparing drafts with sent emails. Mosseri confirmed that while A.I drafts his emails, he manually edits them before sending, noting that out-of-the-box models are 'the average of the internet'.

15
00:04:39.466 --> 00:04:56.042
Mosseri described building 'loops' to train the A.I, which improves its ability to match his tone and values over time. He stated that the A.I tool Muse compares its drafts against the actual emails he sends to learn from the differences, automating the feedback process.

16
00:04:56.392 --> 00:05:17.704
Mosseri also mentioned using a tool called Muse to engage via text and manage email priorities, suggesting simpler interfaces for proactive agents. His workflow highlights a balance between automation and human oversight, where A.I handles the drafting burden but the human ensures the final message aligns with their intent.

17
00:05:20.103 --> 00:05:47.217
JPMorgan Chase faced a significant technology disruption on November sixteen, two-thousand-twenty-six, caused by a software bug in its on-premise hardware systems. The outage delayed transactions and affected banking services, disrupting interactions with other lenders and some Zelle transfers. This incident highlights the bank's central role in the payments ecosystem, where a single failure can ripple across multiple financial institutions.

18
00:05:47.567 --> 00:06:14.106
While the immediate financial impact appears limited, Nikita Kataruka argues that the incident raises concerns about operational resilience, potentially increasing remediation costs and inviting regulatory scrutiny. In a briefing on A.I News and Strategy Daily, Nate B Jones notes a pattern of concurrent outages involving major banks, an FAA ground stop, a United Systems takedown, and global Outlook disruptions.

19
00:06:14.286 --> 00:06:19.065
The simultaneity of these events has sparked speculation about systemic risks.

20
00:06:19.415 --> 00:06:39.639
Jones suggests this pattern could indicate powerful A.I systems pushing against cyber defenses or the surfacing of twenty-five years of legacy software vulnerabilities. However, he clarifies that it is not confirmed that A.I attacks caused all these events; other causes have been discussed for Outlook and airline issues.

21
00:06:39.819 --> 00:06:49.525
Despite the lack of confirmation, the pattern remains concerning for analysts monitoring the intersection of A.I capabilities and critical infrastructure stability.

22
00:06:51.925 --> 00:07:16.927
Jensen Huang, the ousted C E O of Nvidia, has signaled his willingness to return to the role in his first public statement since his resignation. Huang stated he would be 'honored' if allowed to come back, but the Nvidia board has no plans to reinstate him. There is also significant resistance in Congress regarding his return, driven by ongoing questions about export controls and national security.

23
00:07:17.277 --> 00:07:43.091
The SEC is reportedly investigating Nvidia following Huang's departure, which involved an affair with an alleged Chinese spy. This investigation adds legal uncertainty to the company's leadership crisis. Despite the turmoil, Nvidia reported strong financial results, with Q3 fiscal twenty-twenty-seven revenue of one-hundred-fifteen-point-eight billion dollars, up one-hundred-three percent year-over-year.

24
00:07:43.441 --> 00:08:02.897
The combination of a robust financial performance and a leadership vacuum creates a complex situation for Nvidia. While the company continues to dominate the A.I chip market, the political and regulatory scrutiny surrounding Huang's potential return suggests that the path to reinstatement is unlikely.

25
00:08:03.077 --> 00:08:12.250
The focus now shifts to how the interim leadership will navigate the SEC probe and congressional pressure while maintaining the company's growth trajectory.

26
00:08:14.649 --> 00:08:39.161
The A.I Daily Brief: Artificial Intelligence News reports that Anthropic's research system uses a lead agent to plan work and spin up sub-agents in parallel, each with its own task, context, and output format. This architecture represents a shift for A.I-native firms, moving away from static, standing agents toward dynamic structures where specialized sub-agents are created only as needed.

27
00:08:39.511 --> 00:09:04.364
The A.I Daily Brief: Artificial Intelligence News notes that this orchestrator approach beat a single agent by about ninety percent on Anthropic's internal research evaluations. This significant performance gain suggests that breaking down complex tasks and assigning them to temporary, specialized agents yields higher quality results than relying on one generalist model to handle everything.

28
00:09:04.714 --> 00:09:32.064
The A.I Daily Brief host observes that teams are increasingly moving away from interacting with many persistent named agents. Instead, they are using one main orchestrator that figures out what is needed and spins up sub-agents as required. This shift involves moving from persistent rosters to persistent bundles of context, with agents being created dynamically for specific tasks rather than existing as permanent fixtures in the workflow.

29
00:09:34.463 --> 00:09:55.305
The A.I Daily Brief: Artificial Intelligence News reports that A.I-native companies are deploying agents across all organizational functions, not just engineering. For instance, Anthropic's growth marketing team, which was historically a single person, now uses agents to analyze ad performance and generate new ads.

30
00:09:55.656 --> 00:10:18.162
The report also notes that OpenAI research identifies companies like Clay using agents for account management and Basis for customer onboarding. To standardize these workflows, companies like Ramp and Uber maintain shared skills libraries. Uber's library, for example, contains 3,600 skills used over 30,000 times daily.

31
00:10:18.512 --> 00:10:34.960
That's A.I Voices for today. I'm Maya. Thanks for listening.
Follow the show on Spotify, YouTube, Apple Podcasts, Amazon Music, or your favourite podcast platform, and explore the original voices and sources in the episode notes. See you tomorrow.

