The Uncomfortable Truth About Upskilling
Your company is probably planning AI training. Don't bank on it. Here's why, and what you do instead.
Why Corporate Training Fails
The stats are brutal: 90% of corporate training is forgotten within 30 days. The training happens, you attend, you feel productive, and then... nothing sticks. Back to the old way of doing things.
It's not because the training is bad. It's because corporate training treats learning like an event instead of a practice. One-off workshop. Check the box. Move on. That's the opposite of how skills actually stick.
Plus: corporate training is usually generic. It's not designed for your job, your problems, your role. It's designed to satisfy the maximum number of people. Which means it satisfies nobody really well.
And here's the kicker—by the time your company rolls out AI training, you've already fallen behind if you haven't been learning on your own. AI moves fast. Corporate processes move slow. Acemoglu's data shows AI capability doubles roughly every seven months. Corporate training cycles take months to plan and deliver.
What Actually Works
1. Learning by doing
You learn to use AI by using it on real problems. Not sandbox exercises. Real work. Real stakes. The feedback is immediate and visceral. You remember it.
2. Deliberate practice
Not mindless repetition. Deliberate practice means: pick a specific skill, do it with focus, get feedback, adjust, repeat. 30 minutes daily beats eight hours once a year. The compounding effect is real. After 90 days of 30-minute daily practice, you're fluent. After 10 one-day trainings, you remember almost nothing.
3. Communities of practice
Learning alone is slow. Learning with others who are also trying to figure it out? Fast. Slack groups, online communities, peer cohorts. You share what works, what failed, what you learned. You're not trying to solve the problem alone.
4. Personalized paths
Corporate training is one-size-fits-all. Your path should be specific to what you need. A lawyer needs different AI skills than a marketer. A junior analyst needs a different trajectory than a manager. Generic training doesn't acknowledge this. Your learning path should.
What You Do Instead
Design your own curriculum. What specific skill do you need? What AI tool addresses it? Build a 90-day plan around that one thing. Not "become an AI expert." Specific: "Get fluent at using AI to draft and edit code" or "Use AI to research and summarize for case prep."
Make it daily. 30 minutes. Every day. Non-negotiable. Morning coffee, lunch break, before logging off. Make it a habit. Calendar it.
Find your community. Online, work, wherever. Find two or three other people trying to upskill. Share what you're learning. Steal their prompts. Build on their ideas.
Get feedback loops tight. Try something. See if it works. Adjust. Repeat. Weekly reflection: What worked? What didn't? Why? This is the difference between training and learning.
The Real Talk
Don't wait for corporate to make this official or easy. Your company isn't moving as fast as the technology. You're not on the company's timeline. You're on AI's timeline.
The people who get ahead right now aren't waiting for permission or official programs. They're building their own practice, in public or private, and moving fast.
"Training is something done to you. Learning is something you do. One sticks. One doesn't. Pick accordingly."