🤖 Who will win the battle of AI, always think two moves ahead, ...
I listen to 10 hours of podcasts a week, so you don’t have to.
This is the 37th edition of Best 3 Podcasts of the Week 🥉🥈🥇, featuring Lex Friedman, How I Built This, and Marketing Against The Grain.
Fun fact this week… they all share a common theme - AI. Enjoy!
What you need to know
🔮 The future of programming
♟️ Always think two moves ahead
🤖 Who will win the battle of AI?
🔮 The future of programming
🥉 Third place (4 min read vs 3 hours 22 mins listening)
Developers dominate the best job lists. They boast some of the highest rates of job satisfaction, they’re among the highest paid, and they’re expected to be one of the most in demand jobs for the future.
Lex Friedman and Guido van Rossum discussed the later. Guido is the creator of the Python programming language so he has a lot to say on the future of coding. This roundup is high level for non technical folk like myself. If you’re an engineer, you’ll want to listen to the full episode.
What they say
Coding in 2012
Coding has changed in fascinating ways. So much of coding is copy and pasting from Stack Overflow and then adjusting.
I don’t want to talk down to that style of coding because it’s nicely efficient.
Lex Friedman
Coding in 2022
Do you know where coding is going? GitHub Copilot. I use it everyday and it writes a lot of code for me.
It saves me typing. All I have to do is change one word in a line of text then otherwise the code is generated perfectly.
Guido van Rossum
Coding in 2032 - Part I
It’s fascinating to hear that you use GitHub Copilot. Are you worried about the future of automatic code generation?
With this increasing capability, are programmer’s jobs threatened or is there still a significant role for humans?
Lex Friedman
Coding in 2032 - Part II
[Sarcasm] Are programmer’s jobs threatened by the existence of Stack Overflow? I don’t think so. It helps you take care of the boring stuff.
You shouldn’t try to use AI to do something that you have no way of understanding what you’re doing.
A tool like GitHub Copilot is always best when the question you’re asking is ‘Please remind me of how I do this. I could look up how to do it but right now I’ve forgotten the method.’
AI serves that role. It’s like a great assistant. The creative work of deciding what you want the code to do is totally yours.
Guido van Rossum
What I say
Why it matters: Copilot is changing how developers work today, let alone tomorrow.
My biggest adjustment with using Copilot was that instead of writing code, my posture shifted to reviewing code. Rather than a free form solo-coding session I was now in a pair-programming session with myself. I kept having a verbal conversation with myself…
Okay…
Do I like that?
Is that right?
It’s probably good enough.
That will have to be fixed.
Dave Rupert, co-host of the ShopTalk podcast
My take? In 10 years, it’ll become commonplace for humans to dictate the problem and the desired solution. AI will tackle everything in between. Will there be a time where AI becomes the pilot and we become the Copilot? Maybe. Either way, there’s nothing we can do about it. The key takeaway is to embrace AI, not reject it.
Between the lines: You know me. I love steelmanning. Here’s the argument for rejecting AI…
You could argue that existing AI models are a house of cards about to collapse. Think Napster in the early 2000s. Generative AI has come under scrutiny for allegedly violating copyright law. The first class action lawsuit of its kind was filed last month in the US.
Will this stop companies from training their AI models on publicly available data? Will this slow down the rate of innovation and reduce the impact AI has on our lives?
I don’t think so. Why? Because (1) Microsoft owns GitHub and (2) Google is entering the fray with Project Pitchfork.
Microsoft and Google have the equivalent of an unlimited legal budget. Any copyright lawsuit will have to contend with two companies each worth close to the same GDP of Spain. People will try again and again to defend their jobs with lawsuits. This is futile. AI is here to stay whether we like it or not. It’s best to put your energy into adapting to a new norm and a new way of working.
♟️ Always think two moves ahead
🥈 Second place (2 min read vs 1 hour 18 mins listening)
When you know, you know. Some technological shifts are obvious, even 10 years before they happen. The people and companies that start making their moves early are those who turn out to be an ‘overnight success’. Anthony Wood is founder and CEO of Roku. He revealed to Guy Raz how Netflix did exactly this.
What they say
How Netflix preempted streaming
Reed Hastings is very strategic. He knew that many companies had missed transitions in their industry.
He knew his industry was going to transition from DVDs to streaming. He just didn’t know when.
He told me once that he’d allocated say 3% of revenue to work on this idea of streaming in the background so they didn’t miss the transition.
Anthony Wood
What I say
Why it matters: There are a lot of technology shifts that are happening right now. Shifts that can and likely will disrupt your business as it stands today. AI is of course one of them. But there are others to consider depending on your industry.
Reed Hasting allocated budget today for Netflix to remain competitive in a drastically different tomorrow. This is simple, prudent advice that you may consider implementing for your business or you personally.
Between the lines: For entrepreneurs, this is a good lens to think of new startup ideas. Ask yourself ‘What companies will be needed to service a new industry or a new way of doing things?’ Here are a few second order effects of self driving electric cars to get your gears turning:
Self driving electric cars will need a place to get charged and cleaned
Parking garages will need to be repurposed as less people own a car
Road traffic may increase as mobility costs decrease
Over 1 billion gasoline cars will become obsolete and worthless over time
Car insurance, gas stations, and mechanics will see demand decrease
Injury lawyers will be out of a job - this frees up billboard space across the US!
Enjoy this line of thinking? Dig deeper by reading this post by Benedict Evans.
🤖 Who will win the battle of AI?
🥇 First place (4 min read vs 33 mins listening)
In a fleeting moment of greed and excitement, I explored what it would take to make my own version of avatarai.me. Unfortunately, two things stopped me from hacking together an MVP. (1) I can’t code. (2) I realized AI will become a moat for incumbents, not a threat. Most AI first companies will fail. Kipp and Kieran explained why.
What they say
Kipp’s spicy AI take
I’m long AI. It’s going to have a big impact on society. I’m short and skeptical of AI standalone companies. I don’t think they’re going succeed.
One of the things we’ve learned from software is that if you let people use something for free and bundle products together, it’s an effective way to get mass distribution and adoption.
The problem all of these AI apps are going to have is they have real costs so there’s no free version.
Take any of the big SaaS platforms like HubSpot. We could offer a free version of an AI tool because we have the ability to up-sell. We have a high customer lifetime value so we could eat the costs for the free user base in a way that you can’t if you’re a standalone AI tool.
If you’re going to win as a modern day AI tool, you’re going to need to have very specific use cases.
You need to figure out how to have freemium economics. The first AI tools that go freemium will see mass adoption compared to other tools.
Kipp Bodnar
The King of bundling
There’s going to be a bunch of indie builders that create these AI tools but ultimately they’re going to be copied by larger organizations.
Use LEX as an example. They’re trying to build the AI word processor. They have a head start but Google Docs or Microsoft Word are going to go and copy all of those features.
This is what Microsoft did with Slack. More people use Teams than Slack.
Steph Smith
Distribution is key
Canva just did this. They added a freemium text to image AI component to their tool that makes total sense.
This is Steph’s point - distribution is the only thing that matters.
Grammerly could have killed most of the AI writing space if it had bundled a tool because they already have mass distribution.
It makes more sense for a lot of these AI tools to be integrated into a core product set versus being a standalone product.
Kieran Flanagan
What I say
Why it matters: Over 20% of VC dollars now flow into AI startups. We’ll see some huge winners in the space. However, like crypto, many will fail not because of the underlying technology but because of unsustainable distribution costs. AI presents incumbents with a once in a decade opportunity - make the right moves over the next 12 months and lock in their dominance for another 10 years.
Between the lines: Take my AI for Moonpig idea. As a customer, I’d love to send personalized AI generated birthday cards at the push of a button. However, if I were to start this business, the race would be over before it even started.
Moonpig would invariably copy this feature. And that’s the crux of the issue. Most AI tools are features, not standalone businesses. Given most companies build on top of the same open APIs, many products are undifferentiated. An existing brand with strong distribution channels will win 9 times out of 10. With that said, I love an underdog. Don’t let this put you off from being the 1 out of 10!
Shoutouts
When I find newsletters, podcasts, or books worth sharing, I’ll feature them here:
TimeMachiner brings curious readers that next great article by sharing the best in tech, culture, and nostalgia. Why do you want this? Because you’ll get irresistible articles from corners of the internet that you never knew existed. Aaron Crocco shares 3-6 rabbit hole stories ready for you to explore and deep dives into a single topic from untraveled parts of the internet.
Note, these quotes were pulled at different points of the episode. Some sentences were left out to make the narrative clearer and more concise. Podup is not associated or affiliated with any podcast (unless otherwise stated). All roundups are independently written and do not imply any sponsorship or endorsement by the podcast.