I have always been intrigued by the rapid developments in AI. The world is moving at an accelerated speed—something you might not fully notice from Bangladesh. Let me give you a snapshot of what’s happening.

1. AI Efficiency & Cost Reduction #

The efficiency and cost reduction in AI are said to be doubling every four months. At this rate, AI is on track to become commoditized—meaning it will be as accessible as using a gas stove, a mobile phone, or the internet.

For more insights, have a look at Bob McGrew’s interview on Y Combinator’s YouTube channel.

2. AGI is Closer Than We Think #

• Demis Hassabis (DeepMind CEO) predicts AGI in 3-5 years.
• Dario Amodei (Anthropic Co-founder) and Sam Altman (OpenAI Co-founder) share a similar timeline.
• Lex Fridman predicts by 2030.
• My take: Sooner than 2030, but keep in mind that “AGI” might mean different things to different people.

3. The Small Model Breakthrough #

If a small model such as Microsoft Phi could achieve GPT-4 level reasoning using “common sense” data and carefully curated reasoning datasets, it would be a major disruptor to the current foundation model industry. Massive data centers might become unnecessary if these compact, high-performance models become a reality.

Related discussion on Arvind Srinivas x Lex Fridman Podcast.

4. Vertical AI Agents Are the Startup Play #

The AI startup trend is shifting toward Vertical AI Agents—specialized AI tailored for specific industries. Examples include:

• A Voice Agent for designers
• A Research Agent for lawyers

YC’s Request for Startups (RFS) is actively looking for ideas in this domain.

5. Google Referrals Are Down #

Google referrals reportedly dropped 15% in the last year (according to YC research). While not in Google’s official annual report yet, it’s an important trend to keep an eye on.

6. Stack Overflow’s Traffic Collapse #

Stack Overflow’s traffic started declining in 2022, even before ChatGPT launched, primarily due to GitHub Copilot. By now, traffic is down 60% this year, revealing how developer workflows are changing.

7. Economic Shifts & AI’s Impact #

Historically, new technology that dramatically increases wealth creation has led to large societal shifts. For instance, 97% of the world’s population were once farmers, now it’s less than 3%.
AI is the next wave—humans will need to adapt, and swiftly.

8. AI in Healthcare: A Game Changer #

With an ever-growing need for healthcare services, AI can help meet demand where human resources fall short. This opens tremendous opportunity for innovation in diagnostics, patient monitoring, and administrative tasks.

9. The Value of Evaluation Sets #

A YC founder working on an AI agent said:

“The most valuable thing our company built is the eval set.”

High-quality evaluation sets (test suites that verify AI correctness) are crucial for reliable AI applications.

Learn more from the Lightcone Podcast by YC.

10. The Challenge of Internal Search #

No one has truly cracked internal search—effectively searching within personal files (Google Drive, Docs, local directories). It’s vastly different from web indexing and remains an unsolved challenge.

Again, see the Arvind Srinivas x Lex Fridman Podcast for discussions on this topic.

11. The Role of Search in AI Understanding #

To go beyond pattern matching, AI models need search. Without some form of “search” or “retrieval,” we can’t really say they “understand.” In many research circles, “curiosity” = search.

For a fascinating perspective, check out Demis Hassabis x Alex Interview on YouTube.

12. The DeepSeek Moment & AI Cold War #

The DeepSeek moment allegedly has shaken the US AI industry, raising fears of an AI Cold War for global dominance. Given the complexity of inference and compute, AI might become a winner-takes-all game. The first country or entity to solve key infrastructure problems could dominate the global AI market.

Learn more from Lex Fridman’s 5-hour AI Special.

13. AI Job Displacement is Real #

AI is rapidly automating skill-specific jobs, while management roles and product leadership are becoming more valuable than ever. In short:

Knowing what products to build is the critical skill.

This is why Product Managers (PMs) are increasingly in demand.
If you’re curious, Andrew Ng’s article on AI Product Management offers more insights.

Final Thoughts #

AI is accelerating at an unprecedented rate, reshaping industries, jobs, and even geopolitical power structures. The next decade promises to be transformative. If you’re not paying close attention, you might miss one of the biggest waves in technological history.

Have a great day and stay curious!