GenAI is kinda amazing
If you haven’t spent an hour with Chat GPT or a LLM (Lamda, Bard etc), you’ve been missing out. It’s a super easy, personable chat friend, who is 100% interested in what you are. They are like super-smart, ivy-league interns who keep studying every month. They keep track of what you said, following the line of conversation, and get better over time.
This is an incredible unlock, super-power given to anyone who spends 8+ hours in front of a computer screen – browsing, clicking, writing, reviewing, drawing, editing – because this free resource makes it easier to do 30% of our jobs.
If you’re a white collar worker, you need to get good at your job. Average is not okay because the average (free) LLM is better than that. Also, the GenAI applications are machine learning – they have a learning curve that’s faster than yours, insatiable, and open-ended. Namely, GenAI is an eager-beaver intern. . .putting in the hours, getting better at the craft, listening intently, and open to feedback. GenAI is a good hire.
As a Gen-X, great timing
Was born in 1971, same year as Starbucks and Bangladesh, so I’ve been lucky in many ways:
- Got to experience one of the biggest booms in global prosperity 1980-2020
- Saw the life before the Internet
- Got to experience Internet 1.0, 2.0, now 3.0
- Saw life with pretty weak AI (“Alexa, stop”)
- Saw the transformation of Deep learning, transformer, OMG AI
Eric Schmidt (ex-chairman of Google) and Henry Kissinger (ex-Secretary of State under Nixon) wrote a The Age of AI: Our Human Future (affiliate link) here, where they compare the advent of today’s AI with the Renaissance.
Comparing AI to the Renaissance
Yes, BOOM. that’s big talk. Yes, Di Vinci, Galileo, Shakespeare, Dante, Bacon, Copernicus, Kepler, do I need to go on? And that’s just Europe. The idea that this is a step-function change across work, culture, science, philosophy.
Think Joseph Schumpeter on a logarithmic scale. Think, my my my.
GenAI is trained on the corpus of human data, insight, wisdom, logical structuring. Basically, GenAI is training itself on Sapien-level structuring of the world. Basically, we mammalian bipeds are what it is learning from. It goes to reason, that add 20-30+ years of logarithmic compute + training data + algorithms, this thing can do better.
So, I am glad that over the next 30 years – God willing – I will get to see the next Renaissance. I missed the last one.
Crossing the chasm
in school, I talk about Geoffrey Moore’s classic Crossing the Chasm (affiliate link) here. We learn about the technology adoption lifecycle; how the early adopters and mass market look, smell, feel different. Early adopters are willing to suffer through the kludgey interface, problematic bugs, and beta version of everything. Minimally viable product (MVP). Basically, it’s neat – but is clunky and doesn’t work great. It’s still the domain of the nerds and geeks.
The mass market (on the other side of the chasm) is where the money is made, but these folks are risk-averse. They love how good it works, not how neat the technology is. So the trick is to get technology products into the hands of a targeted (often niche) group who will adopt like crazy, then tell their friends. Bowling pin approach.
ChatGPT skipped right over the chasm
The remarkable thing is that ChatGPT just skipped over the chasm, like it was divot in the grass. Nothing to it.
It seems like there is massive, seamless adoption by everyone I know. “The best marketing is a good product.” Yes, this is a wicked good product, and yes, I’ve been spreading word of mouth daily. Why’s it so good?
+ Ease of use
GenAI is more than ChatGPT, I know that. . .yet, ChatGPT is the one I want to talk about. It was so damn user friendly, and useful, that the resistance to it’s adoption was zero. How do you use it? Type or talk as you normally would. How long does it take to access? A few minutes.
It’s easy to use. That’s the point. It didn’t snake it’s way through B2B enterprise applications, then 10 years later find it’s way into Costco. Nope, it’s web-friendly, mobile-friendly. You talk to it like Star Trek, and it is very helpful.
+ Relevant
How relevant is it to you? ChatGPT is 100% interested in what you want to talk about – it’s an infinitely patient tutor, translator, proof-reader, friend, poet, and counselor. It’s crazy.
The context window (fancy word for search bar) can now be infinite. Basically, if the responses are not great – it’s probably your fault. You need to guide this Star Trek device to the right parts of the InterWeb. Get better at prompt engineering. Get better at delegating tasks.
+ Improving
Machine learning is remarkable because – well – it’s constantly learning. The machine learning is voracious and indefatigable; it just gladly learns more and learn. As if “Kaizen” was it’s middle name.
As I often remark, right now, the GenAI is the worst it will ever be.
It takes feedback. Sometimes it gives you two choices (A/B tests) and asks you which level of detail and format you prefer. Like a good junior consultant, it is very coachable.
+ Doing good work
No reason to sugar coat this. ChatGPT adds value when I work with it. Pulls up relevant data. Proof reads slides. Advises on the best PowerPoint titles. Evaluates gaps in my logic. Tells me if the tone of a sentence is odd. Want to explain a difficult concept at a high school level using cooking analogies? ChatGPT has got you covered.
+ Feeding your curiosity
There’s never been a cheaper (money, time, hassle) way to learn. As consultants, our job is to get good at asking questions – with the right framing, in the right order. Question 1 -> Answer 1 -> Question 2 -> Answer 2 -> Question 3 -> Answer 3. Spend 25 minutes digging, reading, learning, asking and you’re getting smarter.
So What?
1. Use the tools
These tools are there for you to use. Everyone else is using them. If you were cooking a great meal, are you such a Luddite that you wouldn’t use a microwave or food processor if it would help you cook good food?
- “Put that information in a table for me”
- “Put dates, products, and price in columns”
- “Redo the itinerary now, but for 10 days with a budget that is 20% less”
2. Get creative
Every week I learn a new use case with ChatGPT. If you were in my office, you would hear me say, “oh wow”, as if I were 10 year-old going to Dave & Busters video arcade for the first time. Mind blown. In Korean, MENG-BOONG.
Don’t compete against the food processor. Just learn to use it in more creative ways.
3. Ask for feedback
Ask ChatGPT for feedback. Did you know that you can take a screenshot of your PPT and add as a photo?
“Evaluate this document on a scale of 1-5 best” and you will marvel how it gives you 3s and 4s. It will have some criteria and give you golden feedback, without the shame of wasting your manager’s time.
Describe the type of audience of your presentation, and ask it for criticisms of where you might get questioned. This is scenario planning on the cheap.
4. Know the limits
Most large language models (LLM) are based off of the Internet. Not a trick question – do we agree that a lot of the content on the web is trash? Yes, of course. Google index hundreds of billions of pages, which means there must be trillions of pages. What could go wrong?
ChatGPT – however good it is – is giving you a super generic, somewhat over-polished, longish answer. (Yes, you can give it more directions and narrow the search, and modify the output style).
The point – don’t rely on it for a finished product. This tool is your intern.
5. Don’t be average
This is a terrible time to be average. GenAI will commoditize the low-end of the white collar market. This is a fancy way of saying, “if you are low-skilled paper pusher, watch out.” This is not meant to be derisive or dismissive. We all have administrative, bureaucratic, low-value tasks on our daily plate.
If you’re spending a lot of time researching, clicking, copying/pasting, summarizing – I just want to say that ChatGPT is good at a lot of those things. Time for you to move “up market” to more difficult, nuanced, innovative tasks. Things that the machine has trouble doing.
Strategy is about creating an unfair advantage. Not competing with ChatGPT.
Of course, there is a hype cycle
As with anything, there is also a hype cycle at work here. Estimates vary of course, (that’s why they are called estimates), but AI-related capex for the top 5 companies will approach $200 Billion this year. That doesn’t even count the multiplier effect downstream to Samsung, Nvidia, TSCM, Applied Materials, ASML, etc. . . money going after this with abandon.
Will some people lose money? Of course, lots of Internet 1.0 went out of business.
That doesn’t mean the the Internet didn’t matter. It just grows, shrinks, helps, and hurts in starts/stops.
How to not be average?
This is a strategy course worth of advice, but a few things come to mind:
- Know your unfair advantage, unique an rarified talents
- Figure our your Venn diagram – what you like, what you’re good at, what people will pay you for
- Set your time frame; are you 3rd career, semi-retired or 22 years old, with 40+ year runway ahead of you
- Are you course-correcting from good feedback?
- Are customers paying you for your work? Have you written a proposal and sold something?
- Solve novel problems with novel approaches; create massive value for customers who are willing to pay
- Develop a safe network of win-win relationships; Identify the clients, patients, customers, people you serve
