Another Crazy Day in AI 7/118

Another Crazy Day in AI: Urgent, Unresolved and Moving Fast

Feb 25
3 min read

Hello, AI Enthusiasts.

Here's another crazy day in AI:

Google DeepMind boss warns AI threats need urgent attention
Anthropic updates safety policy as stakes rise
Perplexity introduces a new AI computer
Some AI tools to try out

🎧 Listen to a quick breakdown of today’s stories.

Urgent, Unresolved and Moving FastAnother Crazy Day In AI: The Podcast

2:18

TODAY'S FEATURED ITEM: DeepMind CEO Calls for Urgent AI Regulation

A robotic scientist in a classic white coat with 'AI Scientist' on its back stands beside a human scientist with 'Human Scientist' on their coat, looking towards the AI Scientist.

Image Credit: Wowza (created with Ideogram)

How prepared are we for the risks that may come with rapidly advancing technology?

Google DeepMind CEO Sir Demis Hassabis sat down with BBC News for an exclusive interview at the AI Impact Summit in Delhi, the largest-ever global gathering of world leaders and tech executives, and didn't hold back. From urgent calls for smarter regulation to candid admissions about the limits of his own influence, Hassabis offered a grounded but sobering look at where we are in the race to develop AI responsibly. The article, reported by BBC Technology Editor Zoe Kleinman and producer Philippa Wain, captures a world at odds over how, or whether, to govern one of the most powerful technologies ever created.

A few points that stood out:

Demis Hassabis said more urgent research is needed to better understand emerging risks.
He highlighted two major concerns: misuse by bad actors and the possibility of losing control as systems grow more capable.
Hassabis called for “smart regulation” and stronger guardrails while noting that Google DeepMind is only one player in a much broader ecosystem.
Sam Altman also voiced support for urgent regulation during the summit.
The U.S. delegation, represented by Michael Kratsios, rejected the idea of global governance, warning against heavy centralized control.
Narendra Modi stressed the importance of countries working together to ensure broad public benefit.
Hassabis said the West is slightly ahead of China in development but suggested the gap could close quickly.
He expects the technology to significantly expand what people can create over the next decade.
Technical education in STEM will remain valuable, though creativity and judgment may become increasingly important.

The summit brought together delegates from over 100 countries, all with different stakes in how AI develops from here. A joint statement is expected before the event wraps up, though how much common ground was actually found in the room is still worth paying attention to.

What the summit did surface, pretty openly, is that there's genuine disagreement at the highest levels on how to move forward with AI. The people building it, the governments watching it, and the countries competing over it are all working from different playbooks — and that gap is something most of us will eventually feel the effects of.

Read the full article here.

Watch the interview here.

OTHER INTERESTING AI HIGHLIGHTS:

Anthropic Updates Safety Policy as Stakes Rise

/Clare Duffy, (Tech Reporter), and Lisa Eadicicco, (Tech Editor), on CNN Business

Anthropic is revising its long-standing AI safety policy, replacing hard commitments with a more flexible, nonbinding framework. The company says the change reflects the rapid pace of competition and the need to iterate, though it arrives amid pressure tied to a potential $200 million Pentagon contract. Critics see the move as a meaningful shift from Anthropic’s earlier safety posture, while the company maintains it is strengthening transparency and accountability. The update underscores growing tension between AI safety ambitions, government interests, and industry competition.

Another Crazy Day in AI: Wall Street's Newest Fear Is a Work of Science Fiction

Feb 24
3 min read

Hello, AI Enthusiasts.

Here's another crazy day in AI:

Jim Cramer on the report that rattled markets
Could Advanced AI slow economic growth?
Lyft speeds up customer support With Claude
Some AI tools to try out

🎧 Listen to a quick breakdown of today’s stories.

Wall Street's Newest Fear Is a Work of Science FictionAnother Crazy Day In AI: The Podcast

3:41

TODAY'S FEATURED ITEM: Jim Cramer Says the Panic Is Overblown

Image Credit: Wowza (created with Ideogram)

What if the very thing the market has been counting on turns out to be its biggest problem?

In a recent episode of Mad Money, host Jim Cramer addressed growing concerns that artificial intelligence could significantly disrupt white-collar employment and the broader economy. The discussion was prompted by a highly pessimistic research paper forecasting widespread job losses and pressure on the middle class. While the market's recent pullback reflects how seriously some investors are taking that scenario, Cramer expressed skepticism that the outcome will be as severe. He acknowledged the technology will affect valuations and business models, but argued the longer-term impact may be more complex than current headlines suggest.

Key developments discussed:

Market declines were partly fueled by worries that automation could reduce demand for white-collar roles.
A widely circulated report projecting a severe 2028 downturn has added to investor unease.
Enterprise software companies such as Salesforce, Workday, Adobe, and ServiceNow were among the biggest decliners.
Cybersecurity stocks also fell after Anthropic indicated plans to expand into the sector.
Cramer questioned whether large-scale job losses will unfold as quickly as some forecasts suggest.
Even with that skepticism, he noted that uncertainty could continue to pressure valuation multiples across tech and services.
Concerns tied to private equity and credit conditions may compound broader market fragility.
He continues to see GE Renova as positioned to benefit from growing power demand.

The paper that triggered the selloff may or may not prove accurate, but the market's reaction to it says something real about where investor sentiment currently stands. When a single speculative report can move indices by that magnitude, it suggests confidence in the current pricing of many stocks is thinner than it looks on the surface.

Cramer's underlying point, that valuations matter just as much as the underlying narrative, is worth keeping in mind. Whether AI turns out to be the economic disruption the paper describes or something far more manageable, the way assets are priced going into that uncertainty is a question investors will be working through for some time.

Watch it on YouTube here.

OTHER INTERESTING AI HIGHLIGHTS:

Could Advanced AI Slow Economic Growth?

/Alex Imas, (Professor at UChicago Booth), on his Ghosts of Electricity newsletter

A new essay examines whether advanced AI could, under certain conditions, actually slow economic growth rather than accelerate it. Using economic models and thought experiments, the analysis explores how large-scale automation might reduce demand if income shifts heavily from workers to capital owners. The author ultimately argues that outright negative growth is unlikely but warns that demand-side pressures could still dampen AI-driven gains. The piece also points to policy tools — such as broader capital ownership — that could help keep growth on track.

Another Crazy Day in AI: What If Your Behavior Was Already Predictable

Feb 18
3 min read

Hello, AI Enthusiasts.

Here's another crazy day in AI:

The startup that simulates human behavior
AI recruiting and bias compliance
Anthropic launches Claude Sonnet 4.6
Some AI tools to try out

🎧 Listen to a quick breakdown of today’s stories.

What If Your Behavior Was Already PredictableAnother Crazy Day In AI: The Podcast

3:56

TODAY'S FEATURED ITEM: How One Startup Is Turning Behavior Into Data

Image Credit: Wowza (created with Ideogram)

Could the way you shop, speak, or make decisions already be predictable enough to simulate?

A recent segment from Bloomberg Technology, hosted by Caroline Hyde and Ed Ludlow, looks at startup Simile, which has raised $100 million to develop behavior-prediction tools for business use. In the interview, co-founder and CEO Joon Park explains that their models are trained on consented life-story interviews, transaction histories, and behavioral research, then used to run large-scale simulations of how different types of people might respond in specific situations. Early users mentioned include organizations such as CVS and Gallup, which are testing simulated customer panels and scenario modeling.

Points raised in the discussion:

The company builds digital agents designed to represent different customer profiles and preference patterns using interview data and behavioral science inputs.
Data gathering includes voice-based interviews and open-ended personal narratives, collected with participant consent and used as training material.
Simulated panels are being used to test product ideas, layouts, and concepts before real-world rollout.
Some enterprise teams are using simulations to prepare for earnings calls, with reported success in anticipating a high percentage of analyst questions.
The technology is positioned as a complement to surveys and focus groups, not necessarily a replacement.
The startup emerged from academic research and moved into commercial pilots within a relatively short development window.

It's worth noting that behavior prediction as a concept has been around for a long time — market research, polling, and consumer analytics have all tried to answer versions of the same question. What's different here is the scale and the method, and whether training on life-story interviews produces meaningfully better results than what came before is still something the industry will need to work out.

The companies testing it seem to find it useful so far, but early adoption doesn't always tell the full story. How accurate these simulations are across different groups of people, and in less controlled scenarios, will probably be the more telling measure down the road.

Watch the conversation here.

OTHER INTERESTING AI HIGHLIGHTS:

AI Recruiting and Bias Compliance

/Caroline Carrier, (Associate in Labor & Employment Practice Group), on Employment Law Worldview Blog

As AI-powered hiring tools become more common, new legal and compliance risks are coming into sharper focus for employers. This piece walks through a major discrimination lawsuit involving automated screening systems and examines how algorithmic decisions can unintentionally reinforce bias. It highlights how courts are starting to test where responsibility falls — on the software provider, the employer, or both. The takeaway is clear: companies using AI in recruitment still need human oversight, transparency, and regular bias checks.