Sunday Reads #193: Prompt engineering... for humans?
"Humans are AGI and we still need prompting from our managers" - Mike Taylor
Hey there! (and to recent subscribers, welcome!)
Hope you’re having a great weekend.
If you missed my last newsletter, here it is: The Ultimate Guide to using ChatGPT.
In that article, I wrote about better prompting for chatbots. This week, let’s look at prompting… for humans?
1. Prompt Engineering… but for humans.
Last week, I wrote about how "prompt engineer" is one of the shortest-lived job titles:
My Twitter feed is full of AI "experts" talking about a crazy new phrase that will 10x your prompting game.
Tell the AI "Try harder. You can do it.".
Tell the AI you'll tip 200 dollars for a perfect answer.
Ask it to "Take a deep breath".
Order it like a military general, rather than as a boss terrified of his Gen Z intern quitting.
Promise it tickets to the Taylor Swift concert in Singapore ¯\_(ツ)_/¯
But here’s the thing:
This magical intonation is not the answer. Neither in the long term, nor in the short term.
In the long-term, well... If you believe AI is on an accelerating tear - which all prompt engineers do - then you must also believe that very soon, AI will understand what you're asking. Even if you ask badly.
When I tweeted this, Mike Taylor (whose Five Pillars of Prompting I shared in the article) retorted in reply:
This is true!
With the right "prompt", you can indeed 10x the effectiveness of your teams.
I wrote about this in The right question is a force multiplier:
The right question is a force multiplier. It creates immense leverage for you as a manager. Ask the right question at the right moment, and you can move mountains.
Ten seconds is all it takes.
Way back in July 2020, I had written about the incredible power of asking the right question:
Questions are far more powerful than answers.
They frame the conversation. You can only answer questions that are asked.
The right question takes you right to the essence of a problem, and helps you solve it.
For example, asking a startup what their defensibility is, is far more useful than asking them about the addressable market size.
And the wrong question can waste years.
“What cool feature should we build next?” is the wrong question, when your onboarding funnel itself is leaky.
You can only answer questions that are asked. Therefore, ask the right questions.
In the article, I also shared my favorite managerial question of all time:
"This person you're complaining about - have you tried talking to them?"
A variation of this is the best managerial question of all (and a question I LOVE to ask):
The article itself has more (and here’s the TL:DR).
In the same vein of better prompting for humans, I loved this article from Wes Kao: Super Specific Feedback: How to give actionable feedback on work output.
Why is feedback important? Because:
Training is the highest leverage activity a manager can do!
Wes quotes Andy Grove in her article:
"Ninety minutes of your time can enhance the quality of your subordinate's work for two weeks, or for some eighty-plus hours."
This quote is directionally correct. But only directionally. Because it understates - hugely - the importance of feedback and training.
I wrote about this in Effective Team Management (even while working remote), again heavily inspired by Andy Grove:
Principle 2: Train your team (all the time!), and give better feedback.
Training and giving feedback are very high leverage activities. In fact, they’re the highest leverage things you can do as a manager.
Why is that?
Let’s say a person in your team is struggling with modelling the financials for a new initiative. You spend 4 hours today walking through that with her. 4 hours = 10% of one work week.
If this improves her performance by even 1% every day, you keep reaping the benefit all year.
And like bank interest, this compounds too.
Learning how to do one task often unlocks a new insight in another related task. Now that she knows how to model financials, she also does cost-benefit analyses better.
She’ll also train her teams and stakeholders tomorrow, in the same way you have.
And all the benefit comes back to you! Remember the Manager’s Equation.
Training is a gift that keeps on giving. It’s like Amway, except it’s not a ponzi scheme.
Every conversation is an opportunity to train. Every decision is an opportunity to train.
So, coach your team. Coach them daily.
Going back to Wes' article, some of the key points I liked:
A. Be more specific than you think you need to be.
I often like to say, "Specificity is a superpower". (Note to myself: should write more about this in the coming weeks).
This applies to giving feedback too. When you make an effort to be clear, it pays incredible downstream dividends.
From Wes' article:
Regular feedback tends to be general, abstract, and mainly about behavior.
Super Specific Feedback is extremely concrete feedback primarily on work output, such as writing, product flows, marketing assets, and design. The goal is to strengthen the work product to get it closer to ship ready, and to help the feedback recipient improve their craft and judgment over time.
As she says later in the article:
Aim to be tactical, actionable, concrete, and specific.
Tactical: “This isn’t a generic, hand-wavy theory.”
Actionable: “I can put this into practice right now and apply it to my future work.”
Concrete: “This can be observed and measured.”
Specific: “This is precise advice for a particular situation.”
Some things that have worked for me:
Always edit in track changes. Microsoft Word and Google Docs make this easy. But even when it's an email draft, I use a red font and strikethroughs to show what I'm editing. And I add comments to explain my reasoning.
Use Loom. Loom is a game-changer for me. It takes a lot of effort to put down structured feedback in an email. And if you want to do it over a call / in-person instead, then you have to schedule the meeting. Instead, record your screen and talk through your changes. 10 min, and you're done!
In fact, because giving feedback becomes easier, you tend to do it more often. And your team (and you, as a result) benefit immensely.
I cannot advocate this enough - if you do knowledge work and manage a team, take a Loom paid subscription. Few things are higher ROI than this.
B. Explain the Why.
It's not enough to say what you want. It's also important to explain why.
Why is this important?
Going back to my Effective Team Management (even while working remote):
Principle 1: Don’t make 100 decisions when one will do.
…
I head business development for a retailer of beauty products. Most days, my team is negotiating with brands that we carry, or brands that we want to carry.
Most negotiation impasses are standard. If the economics aren’t working, it’s due to 3-4 reasons. If the brand doesn’t agree on our purchase targets, there are again 2-3 ways to get on the same page.
When a new team member encounters such problems, I don’t just share the solution. I also make sure to talk through the logic while doing so.
How to identify the underlying problem causing this impasse.
How to resolve that specific type of problem.
How to problem-solve together with the brand.
After a couple of turns through this process, they get it. The next time such a problem occurs, I only find out after it’s solved. Success!
To build leverage, always solve problems at the level of principle.
Remember: Decisions are an opportunity for managers to guide their teams on the right way to do things.
Make the effort to explain context to your team members. Even though you want to feel like Yoda and deal in one-liners.
You might disagree with me. You might say, sorry Jitha, I don’t have time. I’m busy fighting fires right now.
No. You ALWAYS have time to build leverage.
That’s how you prevent the next fire.
C. Tell them what they did well.
Feedback isn't only about what can be better. In fact, "areas of improvement" should be a minority of the feedback.
Instead, focus more on what they did well.
As Wes says:
The ultimate goal for feedback is behavior change. You want to give positive feedback because you want them to continue doing what’s working. Otherwise, your direct report might accidentally stop doing something because they didn’t realize it was effective.
I agree!
I wrote about this in What I learnt from 6 hours with a World Cup winning coach:
Focus on strengths over weaknesses.
Paddy gave an example of his work with a marathoner.
Let's call this runner A.
A always used to come second to another runner B, because B had the ability to get a "second wind" in the last lap of the race. Runner A would lead for most of the race, but then in the last mile, out of nowhere, B would sprint and come first.
So A's coach trained him for a full season, on getting that kick at the end. Result - he did manage to get the last-minute surge. But because he didn't focus on the first 25 miles in training, he came far behind anyway.
When Paddy started working with runner A, he doubled down on his strength. He focused on making what he was already good at - the first 25 miles of the race - 10% better.
The result:
A was able to put so much distance between him and B in the first 25 miles, that B couldn't catch up even with a second wind at the end.
Focus on maximizing your team's strengths, vs. filling the weaknesses.
TL:DR
Training is the highest leverage activity a manager can do. Coach your team. All the time.
Be more specific than you think you need to be. Specificity is a superpower.
Explain the why. Remember: Decisions are an opportunity to guide your team on the right way to do things.
Focus on strengths over weaknesses. First, tell them what they did well.
Link Love:
This is a tweet-sized masterclass in coaching (click here or on the image to see the video on Twitter).
And Kim Scott's Radical Candor 2x2 is a good framework to remember, next time you feel awful about giving direct feedback. (More on it here on First Round Review).
Before we continue, a quick note:
Did a friend forward you this email?
Hi, I’m Jitha. Every Sunday I share ONE key learning from my work in business development and with startups; and ONE (or more) golden nuggets. Subscribe (if you haven’t) and join 1,600+ others who read my newsletter every week (its free!) 👇
2. Golden Nugget of the week.
Saw this chart on the effect of different activities on depression (the more negative on the chart the better).
Not only does exercise beat SSRIs and other meds, but look at Dance!
TIL: We should all dance more! 🕺💃
3. Narrative Violation of the week.
I came across this BBC article recently: New York medical school eliminates tuition after $1bn gift:
A New York City medical school will offer students free tuition following a $1bn donation from the 93-year-old widow of a major Wall Street investor.
The gift to Albert Einstein College of Medicine came from Dr Ruth Gottesman, a former professor at the Bronx school…
In a statement, university dean Dr Yaron Yomer said that the "transformational" gift "radically revolutionizes our ability to continue attracting students who are committed to our mission, not just those who can afford it".
A noble gift 💯. But my first reaction was: Hmm, is that all? 🤔
A whole billion dollars, and all you can do is eliminate school fees? At ONE medical school?
Money - even billions of it - does a lot less than you think, once you scale to a whole population.
Disagree?
OK then, let's do a pop quiz. Quickly, without googling:
If Elon Musk's entire wealth was used to fund the US government's fiscal spend, how many years could it fund?
...
...
...
If you answered with anything higher than 0.03 years (~10 days), you're wrong.
That's right. The entire wealth of the world's richest man would fund only 10 days of the US government's fiscal spend.
[PS. If you want to quibble, ok fine. Bernard Arnault is the world's richest man, and his wealth would last 13 days. Congratulations.]
4. Neuralink!
I saw this video on twitter last week (it's 9 min long, but you can skip through it quite fast):
In the video, the first Neuralink patient (paralyzed from the shoulders down) shows how he's using his Neuralink chip to control his computer... with his mind!
He's moving pieces in an online chess game, by just thinking about it!
A brain-computer interface might sound very "cyborg" to us, but it'll change so many lives.
What's next?
Elon says that Neuralink's next big project will be to help blind people see.
No big deal. Move along, nothing to see here.
5. This is the best opening line of a book I've ever read. 👏👏👏
And it's a book on statistical mechanics, if you'd believe it! Whoever said physicists don’t have a sense of humor…
That’s it for this week. Hope you enjoyed it.
As always, stay safe, healthy and sane, wherever you are.
I’ll see you next week.
Jitha
[A quick request - if you liked today’s newsletter, I’d appreciate it very much if you could forward it to one other person who might find it useful 🙏].