Newsletter #26: AI is enabling computers to learn our language vs. us having to learn theirs. 5 reasons to be stoked on 2024.
I’ve been working on launching my AI with Alec podcast over the last couple of weeks but was stoked to get back to writing.?
The podcast is on its way so without further delay, let’s get back to writing.
Artificial Intelligence is enabling computers to learn our language as opposed to us having to learn their language.?
As a result, one of the biggest constraints to building software - engineers that can code - is being lifted.
To be clear, this doesn’t mean software engineers are no longer needed. Quite the opposite, actually.
It means that coding was how we communicated with machines to build software but now / in the near future, those that don’t know how to code won't necessarily have to work through an engineer to build software.?
Instead, they can go directly to the machine aka democratizing access to building software.?
This is transformative as it fundamentally alters value chains and workflows.
For example, look at the default prompts ChatGPT is serving up...
When you think through the implications of LLMs that can code, there are many exciting benefits to consider, but there are 5 that I’m most excited about.
While there are other, complementary factors connected to LLMs that can code that can be be considered, I think the back and forth between Chamath Palihapitiya and David Friedberg during episode 160 of the All-In Podcast helps put into context how fast things are about to start moving in 2024.
22 minutes, 14 seconds:
“And I think the reason is that we are underestimating how cheap it’s going to be to copy an existing business in 2024."
"And so if you assume that these models (LLMs) are going to get 10 and 100 times better and you assume the cost of compute is going to get 10 and 100 times cheaper and you assume the cost of energy is going to get 10 times cheaper, you’re no longer measuring in decades when a company will be subject to disruption.”
“I think you’re measuring it in, frankly, months.”
“And so I think you’re going to be able to create these companies for very cheap and essentially have them attack an existing business which has upside on economics because they have just a lot of people and a lot of processes that these GPTs can replicate for essentially free.”
32 minutes, 46 seconds:
“But I think these tools to write code, no code tools, copiloting tools and the ability for engineers to be 20, 50, 100 times more productive to build custom applications for their enterprise are so incredibly powerful.”
A few days later, Chamath announced his “8090” incubator to harness LLMs that can code and capitalize on the opportunity to target legacy enterprise SaaS.
With all of that as a backdrop, here’s a bit more detail on why I’m so bullish on LLMs that can code.
1: More / better ideas, prototypes + proof of concepts:
As my man Ron Shaich said in his book “Know What Really Matters”...
“It had never been clearer to me why billion-dollar companies end up in trouble. They had lost their ability to know what matters to their customers, and to use that as a North Star for innovation."
"They were stuck in the past as the world moved forward. Or, to put it another way, they’d lost their connection to discovery while becoming experts in delivery.”
“Discovery and delivery. Those two essential activities must be kept in balance if a company is to remain relevant as it grows."
"Discovery is the leap of faith that brings an innovation to life - the creative work of understanding customers’ needs, seeing what matters, and seizing opportunity."
"Delivery is the work of running the company - making it operational, sustainable, scalable. Both activities are critical."
"Discovery is the lifeblood of the company; delivery is the circulation system that keeps it flowing."
"Discovery is what’s effective; delivery is what’s efficient.”
Empowering better “discovery” while empowering “delivery” to translate their domain expertise directly into ideas, prototypes and proof of concepts feels like the personification of 1 + 1 = 3.
I won’t belabor the point but it’s super hard not to get excited about the kind of tangible innovation that can be unleashed.
*If you missed the 4 minute video of my 6-part sales pitch of why you should buy “Know What Matters” by Ron Shaich, you can find it here. Warning: if you watch the video, you’re going to buy his book… ;)
2: Lower development costs for innovation initiatives:?
Back in my Management / Technology Consulting days, there was no shortage of demand from clients to help them address their perceived challenges in “innovating faster, better and with more tangible results” etc.
But when it came time to talk turkey, there was a healthy tension around the investments vs projected returns that needed to be talked through in order to green light the investments.?
More specifically, investing in innovation “is the leap of faith” as it inherently includes calculated risk taking which ultimately boils down to the probability of projected results being realized aka ROI.
No doubt, the value chain of Idea to Prototype to Proof of Concept becomes substantially more cost effective when LLMs that can code are introduced.
Non-technical leaders can go directly to the machines and convert their ideas into prototypes and ultimately proof of concepts which pulls a lot of costs out of the equation and therefore reduces the investment requirements to see what works.
Simultaneously, one engineer could have the impact of 5, 10, 20…you know the drill on how this impacts the math behind innovation investments.
3: Stronger + faster feedback loops:
Generally, more experiments generate more data which leads to better quality insights.?
Enabling non-technical / semi-technical teams to run rapid prototyping cycles, end-to-end, without requiring access (and approval) to finite technical resources is a big deal.?
Empowering domain experts, those on the frontline interfacing with consumers / clients to drive these innovation cycles with an especially high user-centricity quotient, introduces high quality feedback, faster.
In doing so, introducing LLMs that can code could accelerate the feedback loop between development, users and stakeholders generating a step change in value creation.
4: Reduce ad-hoc / basic requests to free up engineers to focus on higher impact work:
This is a massive benefit that can be underestimated and often overlooked.
Pareto’s principle aka 20% of your efforts drive 80% of your results is real.?
Reducing ad-hoc / basic requests that non-technical leaders can now handle themselves is big because that means that the finite bandwidth of your rockstar engineers can be allocated to solving more complicated, strategic problems to generate more impact for your business.
Good things all around.
5: Improve your company’s Operating System:
The more things change the more they stay the same aka People, Process + Technology.?
When you think about the relationship between the strategic goals of each Business Unit, a centralized vs decentralized tech stack and the teams responsible for “delivery” to realize the company’s ambitions, LLMs that can code are beyond intriguing.?
They can become a powerful engine to evolve a company’s Operating System from 1.0 to 2.0 by automating routine tasks, improving efficiencies, reducing costs, optimizing tech infrastructure and fostering innovation.?
Creating a more lightweight, agile and adaptive operating model that can learn / respond faster and more effectively to rapidly changing market dynamics.
One more thing…it’s hard not to acknowledge the transformative potential of LLMs that can code but change is hard and it includes overcoming valid objections and concerns.
Some might argue that democratizing access brings quality and reliability concerns, loss of specialization, education and training concerns etc.
It’s important to acknowledge and work through these concerns and remember that with proper oversight, education and responsible use, many of the challenges can be mitigated.
The goal is to strike a balance between democratizing access to coding and maintaining the quality, security and standards expected by each organization.
It’s a process but the promise in enhancing both “discovery and delivery” carries benefits that justify overcoming the challenges, even if that means crawling, walking and then running your way into it.
We’ve been here before with transformative technology driving benefits that included those generated by democratization - personal computing, internet access, open-source software, cloud computing, social media, 3D printing etc.
?
While there are challenges to navigate, the benefits are most definitely worth it.
That’s it for this week. I hope you enjoyed it.
Keep an eye out for the AI with Alec podcast. It’s coming soon!
Have a great Sunday and talk soon.
Alec
Data & Tech Career Coach | Confidence Builder | Startup Advisor
10 个月Good luck with the podcast! Excited to tune into it ??