The AI Revolution: Still in the First Inning

We are at the equivalent of 1995 for the internet. The transformative AI era is just beginning.

1. Where Things Currently Stand

Naming specific models here would be wrong within a quarter, so the current line-up lives on the course Tool Radar, which carries a "last verified" date. What is worth understanding is the shape of the capability, which moves far more slowly than the version numbers:

• Frontier models are close to each other. The gap between the leading options is much smaller than the marketing suggests, and the ranking changes every few months. Pick on fit, not on a benchmark.
• Reasoning improved sharply. Multi-step problems that failed reliably a few years ago now mostly work — but "mostly" is doing real work in that sentence, and verification still matters.
• Multimodal is the default. Text, image, audio and video in one model is no longer a distinguishing feature.
• Agents work, supervised. They can use tools and complete multi-step tasks. They are not yet reliable enough to run unsupervised on anything that matters.
• Cost per token has fallen dramatically and keeps falling, which changes what is economic to automate at volume.

2. What Is Probably Next

Predictions with dates attached age badly, so treat these as directions rather than a timetable:

Near-term (12-18 months):

• AI agents will go mainstream: Autonomous agents that can complete entire workflows
• Personal AI assistants: Every professional will have an AI that knows their work deeply
• Real-time voice translation: Flawless, instant translation during calls
• AI code generation: 50%+ of enterprise code will be AI-generated
• Multimodal documents: AI that can read, watch, and listen to entire courses

Medium-term (2-3 years):

• AI tutors: Personalized AI that learns your learning style
• Robotaxis and autonomous vehicles: Full Level 4 autonomy in specific cities
• AI scientists: AI that can conduct novel research
• Brain-computer interfaces: Neuralink and competitors for cognitive enhancement
• Digital twins at scale: Entire organizations simulated in AI

Long-term (5-10 years):

• AGI (Artificial General Intelligence): The debate continues—some predict 2027, others say 2040
• Robotics: AI-powered robots doing physical labor
• Healthcare revolution: AI diagnosing and treatment planning at superhuman level

3. The Skills That Will Matter

As AI handles more tasks, what humans need to excel at:

• Critical thinking: Evaluating AI outputs
• Emotional intelligence: Relationship building, negotiation, leadership
• Creativity with constraints: Novel problem-solving within real limitations
• Systems thinking: Understanding complex interdependencies
• Ethical reasoning: Navigating ambiguity with moral clarity
• Teaching and mentoring: Helping others learn and grow
• Curiosity and learning: Adaptability as the only constant

4. The Skills AI Will Commoditize

These will become baseline expectations:

• Basic coding and no-code tools
• Data analysis and visualization
• Standard content creation
• Research and summarization
• Email and administrative tasks
• Basic design work

5. Preparing for Change

Action steps:

1. Build your learning muscle: Spend 2 hours/week learning something new
2. Stay current without drowning: Use curated newsletters (Ben's Bites, The Rundown, Latent Space)
3. Experiment monthly: Try one new AI tool or technique per month
4. Build your network: Connect with people exploring the frontier
5. Invest in human skills: The more AI commoditizes technical skills, the more human skills appreciate

⚠️ Common Mistakes

• Predicting too specifically: We know direction, not timeline or specifics
• Paralysis by analysis: Waiting for "the right time" to learn AI
• Techno-pessimism: "AI is overhyped"—history shows hype often underestimates long-term impact
• Techno-utopianism: "AI will solve everything"—real change is incremental and uneven

What's Next: How to build a sustainable career in the AI age.