Staying Current with New AI Tools
Lesson 5.6 — Staying Current with AI
The AI landscape moves faster than almost any technology in history. Models that were cutting-edge six months ago are routine today. Tools that didn't exist at the start of the year are now used by millions of people. Keeping up can feel like an obligation — another thing to be anxious about rather than interested in.
This lesson is designed to make staying current feel manageable rather than overwhelming. We'll cover the specific newsletters, YouTube channels, and podcasts worth your time (with honest descriptions of who each is for), a 10-minute framework for evaluating any new AI tool, and a simple weekly habit that keeps you genuinely informed without dominating your schedule.
The Principle: Depth Over Volume
The biggest mistake people make with AI news is following too many sources. More input does not mean better understanding — it usually means anxiety and noise. The goal is a small number of genuinely trustworthy sources that you actually engage with, rather than twenty feeds you skim with mounting guilt.
For most people, two newsletters, one podcast, and one YouTube channel is enough.
Key takeaway: You do not need to follow everything. You need to understand the underlying concepts well enough to evaluate new developments when they arrive — and to know when something genuinely changes the picture.
Four Newsletters Worth Reading
1. The Rundown AI
Who it's for: Beginners and generalists who want a daily briefing without deep technical content Format: Short daily email, 3–5 minute read What it covers: New tool launches, major model releases, business applications, industry news Sample topics: "Anthropic releases Claude 3.7 with extended reasoning"; "OpenAI adds memory across all GPT conversations"; "How a law firm cut research time by 60% using AI" Verdict: The best starting point if you want a daily AI pulse without overwhelm. Consistently well-curated.
2. Import AI (Jack Clark)
Who it's for: People who want to understand what is happening at the research frontier, not just product launches Format: Weekly long-form newsletter What it covers: New research papers, technical developments, policy and safety implications Sample topics: Breakdowns of new model architectures, analysis of AI safety research, geopolitical AI competition Verdict: Dense and technical in places, but the analysis is excellent. Worth the effort if you want to understand why things are happening, not just what.
3. Ben's Bites
Who it's for: Professionals interested in AI business applications and new products Format: Daily, concise, with brief commentary What it covers: Product launches, funding news, use cases, tools Sample topics: New Notion AI features; funding round for AI startup; how a marketing team is using AI for campaign ideation Verdict: Good signal-to-noise ratio. Less hype than many AI newsletters.
4. MIT Technology Review AI section
Who it's for: People who want thoughtful, slower journalism rather than a news feed Format: Weekly digest of MIT Tech Review AI articles What it covers: In-depth features, investigative pieces, long-term trends Sample topics: Investigative pieces on AI and misinformation; analysis of AI regulation in different countries; profiles of researchers Verdict: Slower pace but higher quality on complex topics. Good counterweight to the daily excitement of other AI sources.
Three YouTube Channels
1. 3Blue1Brown
Who it's for: Anyone who wants to genuinely understand how AI and machine learning work at a conceptual level Style: Beautiful animated explainers, rigorous but accessible mathematics Best video to start: "But what is a neural network?" — one of the most-watched explainers on YouTube for good reason Verdict: The single best resource for building real conceptual foundations. Not about news or tools — about understanding.
2. Matt Wolfe
Who it's for: Practitioners who want to know about new AI tools and how to use them Style: Regular "what's new in AI this week" roundups plus tutorials What it covers: New image, video, audio, and text AI tools; practical demonstrations Verdict: High production quality, genuinely useful for discovering tools. Can lean toward enthusiasm over critique, so watch with that in mind.
3. Andrej Karpathy
Who it's for: People comfortable with technical content who want to learn from one of the leading AI researchers and engineers Style: Long-form lectures and technical deep dives Best content: His "Neural Networks: Zero to Hero" series is the best practical AI education available on YouTube Verdict: Genuinely exceptional. Challenging, but if you can follow it you will understand AI far more deeply than most people who use it daily.
Four Podcasts
1. Lex Fridman Podcast
Who it's for: Anyone interested in long-form conversations with AI researchers, entrepreneurs, and thinkers Format: 2–4 hour interviews Sample guests: Sam Altman, Yann LeCun, Geoffrey Hinton, Elon Musk Verdict: Long but rewarding. The best episodes provide rare depth on how the people building AI think about what they are doing. Skip episodes where the guest is not relevant to AI.
2. The TWIML AI Podcast (This Week in Machine Learning)
Who it's for: Practitioners and technically curious people Format: 45–60 minute interviews with researchers and engineers Sample topics: New model architectures, applied AI in specific industries, research frontier Verdict: More consistent quality than Lex Fridman; less famous guests but more reliably focused on what you came for.
3. Hard Fork (New York Times)
Who it's for: General tech-and-AI audience who want context and analysis with some personality Format: 45–60 minutes, weekly What it covers: Major tech and AI stories of the week, with context and occasionally pushback Verdict: Good for understanding the societal and business implications of AI stories, not just the technology. Well-produced.
4. Practical AI
Who it's for: Professionals who want applied, practical coverage rather than hype or research Format: 30–45 minutes, weekly What it covers: How AI is being used in real organisations, tool comparisons, practical guidance Verdict: Underrated. Solid production, genuinely useful focus on application rather than speculation.
Curated Diet Table
| If you have... | Recommend |
|---|---|
| 5 minutes/day | The Rundown AI newsletter |
| 30 minutes/week | Ben's Bites + one episode of Hard Fork |
| 2 hours/week | Add Practical AI podcast + one 3Blue1Brown video |
| 4+ hours/week | Add Import AI newsletter + TWIML podcast |
| Deep learning interest | Andrej Karpathy YouTube series |
| Research/policy interest | MIT Technology Review + Import AI |
The 10-Minute Tool Evaluation Framework
New AI tools appear daily. Most are not worth your time. Here is a process for deciding quickly:
Minute 1–2: The basics
- What does it do in one sentence?
- Who is it for?
- Is there a free tier or trial?
Minute 3–4: The claim check
- What specific problem does it solve?
- Is that problem relevant to you?
- Is this actually better than what you currently use, or just different?
Minute 5–7: A quick test
- Try the most relevant feature with a real task from your own work
- Do not use their demo data or their example — use yours
Minute 8–9: The catch check
- What are the pricing limits on the free tier?
- Where is the data stored and how is it used?
- Is there a privacy policy you should skim?
Minute 10: The decision
- Will this meaningfully save time or improve quality on something I do regularly?
- If yes: add to your regular workflow and give it a genuine two-week trial
- If no: close the tab without guilt
Most tools will not pass this framework for most people. That is the right outcome. Saying no to most tools is how you avoid tool fatigue and stay focused on what actually works.
The Weekly 30-Minute AI Habit
This is the sustainable practice that keeps you genuinely informed without consuming your life:
15 minutes — Scan your curated newsletters Read the items flagged as relevant to your situation. Skip anything that isn't.
10 minutes — Try one new thing Apply one idea from what you've read to an actual task. Not thinking about it — doing it.
5 minutes — Note what changed Keep a simple running document: date, what you tried, what worked, what didn't. This becomes a personal AI capability log that is surprisingly useful six months later.
That is it. Thirty minutes, once a week, and you will be genuinely more capable than the vast majority of people who consume AI content passively without practicing.
Practice Task
This week, subscribe to one newsletter from this lesson's list — whichever matches your current level. Read every edition for two weeks before deciding whether to keep it. At the end of two weeks, assess: does this source consistently tell me things that change how I think or work? If not, try a different one.
You've finished all the lessons in Module 5. Take the quiz to test your knowledge →