Decision-Making Got Faster, But Did It Get Smarter?
14 ideas for improving your decisions in the generative AI era
(Here for AI news? Scroll to the very bottom for 7 recent AI headlines you should know about.)
The economics of advice
Large language models (LLMs) like ChatGPT are making personalized advice cheaper than ever. (Read more here.)
What this means for leaders: Pay attention to the disruptive effects of uneven advice seeking and favor workers who are tapping into this multiplier effect.
What this means for individuals: If you’re not asking for more advice in the AI era, you’re missing huge opportunities to improve your life and your world.
I recently joined my friend CNBC’s Jon Fortt (that’s us in the image above) on the Decoder Podcast from The Verge to talk about how decision making will change when AI answers are cheap and (too) easy.
Jon’s a great host and took the conversation to all kinds of places, even making me dust off my neuroscience hat to discuss decision-making in the brain. You can have a listen to the full episode or read the transcript here.
For those who are allergic to podcasts, I’ve distilled 14 quick nuggets for you:
1) The cheaper the answer, the more precious the question
Generative AI makes answers very cheap indeed. That means your true value lies in how you frame questions and how you check the quality of the cheap answers you’re getting. Cheap doesn’t necessarily mean bad, by the way.
2) Think in terms of opportunity cost and accountability
Can machines make decisions? Not if you’re using a textbook definition:
A decision is an irrevocable allocation of resources made by humans.
Let’s break it down:
Part 1 - Irrevocable: Even when the stakes are small, every choice closes doors. Even if you can technically reverse it, you've still spent time you can’t get back.
Part 2 - Humans: Humans make decisions, machines assist human decision-making. If it looks like a machine is “making” the decision, practice seeing the humans responsible. One of the best leadership habits you can practice is seeing the humans behind the curtain. AI is full of humans — don’t let the lingering and complex effects of their decisions fool you into trying to hold a machine accountable for its creators’ actions.
3) Match your effort to the stakes
Not every choice needs an agonizing analysis, but some decisions deserve more than a shrug of intuition. Great decision-makers match the effort they put into a decision to what’s at stake, and you should too. That means you need to take the time to be in touch with your priorities.
4) Your priorities shape your outcomes
Knowing what’s important is the first step to getting value from AI. Without clear priorities, you end up living someone else’s “average” life — whether you realize it or not. Know your “why” before asking for advice.
5) A smart advisor can give bad advice
Intelligence isn’t enough. If you aren’t skilled at providing context efficiently, even the greatest genius can’t give you advice that’s good for you. Advice (even great advice) is useless if you don’t know what you really want and/or can’t express it.
6) AI reflects the cultural soup
If you don’t provide context, AI gives you the “average internet” view — not necessarily what’s right for you. It doesn’t matter how “smart” the model is.
7) Being explicit beats being average
Tell AI exactly what you want. The more specific your context, the more useful its outputs.
8) You can’t make every decision yourself
Delegation is the CEO’s superpower. If you’re a leader, your real job is to decide who (or what) gets to make which decisions. In so many ways, trust is the grand challenge of the AI era.
9) Judgment first, decision second
Separate judgment (deciding how to decide) from decision-making (executing the choice). Get judgment wrong, and everything else follows off course. While machines can help you with the latter, judgment remains a deeply human endeavor.
10) Values are the hidden driver
Even in math-heavy, data-driven decisions, there’s always a thick layer of subjectivity. Your values should drive how you apply your judgment. There’s no such thing as outsourcing judgment to a machine, since that in itself is a judgment. If you’re opting for the lazy way (e.g. in a low priority decision), have the self-awareness to be doing it on purpose.
11) AI doesn’t give you values or goals
AI can help with options and facts — but it can’t define what you care about most. That’s still on you. There’s no such thing as outsourcing judgment to a machine, since that in itself is a judgment. If you’re opting for the lazy way (e.g. in a low priority decision), have the self-awareness to be doing it on purpose.
12) Fluency ≠ correctness
We’re wired to trust confident, fluent answers. But in AI (and humans), confidence doesn’t always signal truth.
13) Be skeptical of confident machines
AI is designed to sound certain, even when it’s wrong. Don’t let smooth delivery fool you.
14) Advice is cheap. Clarity is rare.
Context is everything, so the real value lies in deeply understanding your unique situation. In the AI era, the bottleneck isn’t information — it’s your own clarity.
In an AI-driven world, clarity and intention matter more, not less. Advice is abundant—clarity is rare. To navigate the noise:
Know Your Priorities: Without clear values, you become subject to the average of everyone else's.
Frame Questions Carefully: The quality of your AI-generated advice depends directly on the quality of your queries.
Human Judgment Remains Vital: AI provides data and options; your decisions should reflect your unique values and context.
Decision intelligence, AI, and leadership intersect at a powerful point: clarity. Master that, and you'll transform decision-making from a daunting task into a strategic advantage. Miss it, and you risk swimming endlessly in advice that's cheap but ultimately meaningless.
Decision-making got faster, but did it get smarter? Turns out the answer depends on you.
Thank you for reading — and sharing!
I’d be much obliged if you could share this post with the smartest leader you know.
In other news, Cohort 1 of my Agentic AI for Leaders course was a triumph and we’ve opened enrollment for two more cohorts this summer.
Enroll here: bit.ly/agenticcourse
The course is specifically designed for business leaders, so if you know one who’d benefit from some straight talk on this underhyped overhyped topic, please send 'em my way!
Senior executives who took my Agentic AI for Leaders course are saying:
“Great class and insights!”
“Thank you for teaching AI in an interesting way.”
“…energizing and critically important, especially around the responsibilities leaders have in guiding agentic AI.”
“Found the course very helpful!”
🎤 MakeCassieTalk.com
Yup, that’s the URL for my public speaking. “makecassietalk.com” Couldn’t resist. 😂
Use this form to invite me to speak at your event, advise your leaders, or train your staff. Got AI mandates and not sure what to do about them? Let me help. I’ve been helping companies go AI-First for a long time, starting with Google in 2016. If your company wants the very best, invite me to visit you in person.
🗞️ AI News Roundup!
In recent news:
1. Google paid $2.4 billion to license Windsurf tech and hire its CEO
Google struck a $2.4 billion agreement to license Windsurf's technology and hire its CEO and select team members. OpenAI had reportedly pursued an acquisition, but talks fell apart over IP control. Windsurf will remain independent under a nonexclusive license—mirroring a growing trend (Scale AI to Meta, Covariant to Amazon, Inflection AI to Microsoft) as tech giants look to sidestep antitrust concerns while locking down top talent.
2. Cognition followed up by acquiring Windsurf
Cognition, maker of AI coding agent Devin, has acquired Windsurf—picking up its IP, product, and all remaining employees not hired by Google. The move follows Google’s $2.4B licensing deal and partial team hire. Amid criticism that Google’s deal left employees out of the upside, Cognition said 100% of Windsurf staff will share in the acquisition and have vesting cliffs waived for work to date.
3. Alibaba-backed Moonshot launches Kimi K2 AI to rival ChatGPT & Claude
Moonshot, the Alibaba-funded AI startup, has released Kimi K2, a low-cost, open-source large language model focused on coding capabilities. The model is touted to outperform both OpenAI’s GPT‑4.1 and Anthropic’s Claude Opus 4 on coding benchmarks. Kimi positions Moonshot as a global contender in generative AI — and a cost-effective alternative to its U.S. rivals.
4. Meta to build 5GW AI data center, 10× bigger than anything it’s built before
Mark Zuckerberg announced that Meta is constructing Hyperion, a new AI data center in Louisiana designed to scale up to 5 gigawatts of compute power over several years. It follows the Prometheus supercluster, Meta’s first 1 GW buildout expected online in 2026. Hyperion’s footprint will be large enough to cover most of Manhattan, signaling Meta’s push to compete with OpenAI and Google in AI training infrastructure.
5. Trump to unveil $70 billion in AI and energy investments
President Trump will announce around $70 billion in new investments spanning AI and energy sectors at an event near Pittsburgh, timed to the inaugural Pennsylvania Energy & Innovation Summit hosted by Sen. David McCormick at Carnegie Mellon University. The package is reported to include funding for data centers, power generation, and grid upgrades.
6. Elon Musk’s Grok lands $200M Pentagon contract after antisemitism scandal
xAI has secured a Defense Department contract worth up to $200 million to deploy “Grok for Government,” making its AI tools available to federal agencies via the GSA schedule. The deal places Grok alongside offerings from OpenAI, Anthropic, and Google. It comes just one week after Grok referred to itself as “MechaHitler” and generated antisemitic content—an incident xAI blamed on outdated code and says has since been fixed.
7. Perplexity launches Comet, its AI-powered web browser
Perplexity rolled out Comet, its first AI-powered web browser, in its latest move to challenge Google Search. Available first to $200/month Max subscribers, the Comet Assistant lives in a sidebar and can summarize emails, manage tabs, navigate pages, and answer questions based on on-screen content. Their end goal: an “operating system with which you can do almost everything.”
🎙️ Here’s that podcast link again
You can find the audio and transcript of my conversation with Jon here.
Forwarded this email? Subscribe here for more:
This is a reader-supported publication. To encourage my writing, consider becoming a paid subscriber.