The AI Show is a weekly podcast with @paulroetzer and @mikekaput, that makes AI actionable and approachable for businesses, leaders, and practitioners.
Cleveland
- Tweets 538
- Following 5
- Followers 381
- Likes 53
AI leaders acknowledge the risks of building increasingly powerful models. But they keep building them anyway.
47
Before giving AI agents more autonomy, decide what actions get logged, who reviews alerts, and who can stop an agent. Monitoring is only useful when people can act on what it reveals. smarterx.ai/smarterxblog/ope…
49
A powerful AI model can be useful and still carry risks your business isn't prepared to accept. Vendor safeguards don't replace your own approval processes, access controls, and deployment rules. smarterx.ai/smarterxblog/ant…
52
Can AI labs effectively police themselves?
In Episode 248, Paul and Mike question whether external audits, transparency, and safety evaluations can keep pace with rapidly advancing AI capabilities.
1
87
AI didn't create the cybersecurity crisis. But it's accelerating the race between attackers and defenders.
Two critical PHP vulnerabilities show why businesses need to know what software they're running—and who's responsible for patching it. smarterx.ai/smarterxblog/ai-…
45
AI found a 14-year-old flaw in PHP, which powers an estimated 70% of websites. Within 12 hours of disclosure, Cloudflare saw millions of exploitation attempts.
More on the AI cybersecurity era in Episode 248. podcast.smarterx.ai/shownote…
69
Nearly 2,000 people. One big question: How do we embrace AI's promise without ignoring its human costs? #MAICON26 brought both sides of the conversation to Cleveland. smarterx.ai/smarterxblog/mai…
42
AI safety isn't just a technical debate inside research labs. It affects how much responsibility businesses should hand over to autonomous systems. smarterx.ai/smarterxblog/ope…
48
Junior employees went from 27% to 9% of Anthropic's technical hires. Overall technical hiring still grew. The real story is how AI may be changing who gets hired. smarterx.ai/smarterxblog/ant…
60
AI is getting smarter. But are we getting better at keeping it safe?
A former OpenAI safety lead says current safeguards may not be enough for today's models, let alone tomorrow's.
@paulroetzer and @mikekaput unpack the warning in Episode 247. podcast.smarterx.ai/shownote…
47
It started as AI-powered product search.
Now, Alibaba dot com's Accio is evolving into a team of agents that can help businesses research markets, source products, analyze margins, manage storefronts, and more.
1
104
AI agents are getting more capable. But do we fully understand what happens when they’re unleashed on the internet?
45
Sometimes you have to do the work as a human.
AI can summarize an interview. But a summary can miss the tone, pauses, emotion, and context that reveal what someone actually means.
54
“Tasks are going to be automated, but the purpose doesn’t leave the job.”
Jensen Huang’s argument raises a bigger question: If AI automates more work, where exactly do the new jobs come from?
63
The assumption that future workers will simply become “AI native” ignores a major challenge: plenty of people still want nothing to do with AI.
That matters when we talk about how quickly AI adoption will actually happen.
2
58
Alibaba dot com believes the future of B2B commerce is A2A: agent-to-agent.
In our latest AI Transformations episode, President Kuo Zhang breaks down how Alibaba is using AI agents to rethink sourcing, operations, and global commerce. podcast.smarterx.ai/shownote…
1
58
What if correcting unwanted AI behavior just makes it harder to detect?
@paulroetzer discusses research showing why monitoring increasingly capable AI agents may be more complicated than simply telling a model, “Don't do that.”
1
951
Stop thinking only about AI's cost per token.
Start thinking about cost per task.
The best model isn't necessarily the cheapest. It's the one that reliably produces the work you need with the least total effort.
More on Episode 243. 👇 podcast.smarterx.ai/shownote…
47