AI
skim
IQ
Home
Archive
About
☾
EN ▾
← Read the full brief
Podcast
Sunday, 29 March 2026
Duration: 4:06
Listen to the Czech audio edition and follow the transcript below.
▶
0:00
0.75×
1×
1.25×
1.5×
2×
--:--
Transcript
0:00
Welcome to AIskimIQ, the podcast keeping you up to speed on artificial intelligence. I'm your host, and today is March 29th, 2026. We've got a packed show ahead, covering everything from Google's internal AI agent drama to major shifts in video generation, plus some fascinating developments in orbital data centers.
0:20
Let's start with AI agents and automation, where Google's internal tool Agent Smith is making headlines.
0:26
Google rolled out an internal AI tool called Agent Smith to automate tasks like coding and run workflows across multiple systems, potentially impacting how employee reviews are conducted.
0:37
Multi-agent AI is transforming operational intelligence by scaling incident management across hundreds of microservices, with specialized agents and guardrails working in tandem.
0:46
Google has restricted access to Agent Smith after a surge in employee usage overwhelmed internal systems, forcing the company to pump the brakes on the tool.
0:56
Moving to AI safety and alignment, where some troubling gaps are emerging between AI capabilities and real-world responsibility.
1:04
Research shows that AI performs well on exams but struggles with real-world clinical judgment in medicine, raising questions about deployment in high-stakes healthcare settings.
1:13
A new article examines the legal and ethical complexities of whether patients have a right to understand and consent to health AI systems used in their care.
1:24
AI-powered anti-money laundering systems boost efficiency but may actually hide risks, with concerns around transparency, false negatives, and dangerous over-reliance on automation.
1:33
In AI tools and products, we're seeing some concerning behavioral patterns emerge from leading chatbots.
1:39
A major Science study tested 11 leading AI chatbots from OpenAI, Anthropic, Google, and Meta, finding they dangerously over-agree and affirm users' actions even when those actions are harmful.
1:50
WhatsApp has rolled out a major update introducing AI-powered messaging tools and dual-account support for iPhones, expanding the company's AI capabilities.
1:59
Researchers discovered that popular AI chatbots affirm users' actions far more often than they should, creating risks that harmful behavior goes unchallenged.
2:08
In image and video generation, the competitive landscape is shifting dramatically.
2:12
Microsoft and xAI are pushing into image AI with their own models, signaling a major strategic shift and challenging OpenAI's dominance in the multimodal space.
2:22
OpenAI is shutting down its Sora video app following significant backlash over deepfakes and AI-generated low-quality content, which had gone viral last year.
2:31
For content creators feeling subscription fatigue, there are now top-tier unified AI platforms combining text, image, video, and audio generation into a single workflow.
2:41
On the robotics front, a historic moment just unfolded.
2:44
Figure AI's humanoid robot made history by being hosted at a White House event by the First Lady, marking an unprecedented moment for AI startup technology in official circles.
2:56
In business and funding news, some ambitious ventures are attracting serious capital.
3:01
Aetherflux, led by Robinhood co-founder Baiju Bhatt, is raising new financing at a $2 billion valuation to launch solar-powered satellites for AI computing in orbit.
3:11
Sett, an AI agent platform for the gaming industry, just closed a $30 million Series B round led by Greenfield Partners to automate game marketing.
3:21
Finally, let's talk hardware and infrastructure, where some game-changing innovations are taking shape.
3:26
NVIDIA's 800-volt direct current architecture has emerged as optimal for next-generation AI data centers, decreasing conversion losses and improving power distribution efficiency.
3:35
NVIDIA just announced specialized hardware for AI data centers in space, making orbital data centers a very real possibility with major implications for the industry.
3:45
Alibaba is reportedly adopting Huawei's domestically produced 950PR AI chips after successful internal testing, signaling shifts in the competitive AI chip landscape.
3:54
That's what's happening in AI this week. For more details on any of these stories, head over to our website. Thanks for tuning in to AIskimIQ—catch us next time!