An AI CEO finally said something honest...

A few days back, I ran into a Reddit

post claiming that an AI CEO finally

said something honest, and we really

need to talk about it. On Valentine's

Day, Dex Rod posted on Twitter his

thoughts on AI code generation. I

believe his take is completely on point.

So, in this Monday morning review, we'll

discuss what the creator of Open Code

Agent has to say about the crazy times

the software world is going through.

Bear in mind that this tweet comes right

around the time when the head of Claude

Code announced that coding is solved. In

a recent podcast, Boris Churnney went

through pretty much all the talking

points we are used to by now. Developers

rely fully on clo. They haven't touched

code since last November. Agents are

running autonomously all over the place.

Productivity per engineer has increased

200%. And in a year or so, knowing to

code will not matter because everybody

will be able to do it. What's funny is

that these statements are coming from

the man in charge of the Cloud Code

GitHub repository, which at the moment

has more than 6,000 open issues. But

remember guys, coding is solved.

Meanwhile, Anthropic is giving $50,000

signing bonuses trying to hire software

engineers. Anyway, I know we are all

sick and tired of this narrative by now,

and this is why Dex's tweet is such a

breath of fresh air. Everybody is

talking about their teams like they were

at the peak of efficiency, and the

bottleneck was the ability to produce

code. This is a key observation because

all the focus at the moment is on the

number of lines produced by the new

model released this week. But like I've

been saying on this channel for the past

year or so, building software products

is much more than writing the code and

the code was never the biggest issue in

this equation. We have had frameworks

like Laravel or Rails for years which

are doing so much of the dev work under

the hood that most of the hard parts of

wiring things together have been

abstracted away for more than a decade.

Putting in place an MVP took a few weeks

at most. So I doubt that anything

meaningful will change if we go from

weeks to days. On top of that, no code

tools have been around for years as

well, and there was no dent in developer

demand. If removing code was the silver

bullet, the industry would have already

shrunk because the real issue was never

writing code. In Dex's words, your

organization rarely has good ideas, so

ideas being expensive to implement was

actually helping. The time constraint of

having a new feature deployed was a good

thing because it forced teams to

prioritize, polish, and cut. When

implementation is expensive, you are

forced to think. You debate trade-offs.

You question whether the feature is

worth the maintenance cost. And you kill

mediocre ideas early because every

mediocre idea has a real opportunity

cost. This hostile environment helped

build the products we know and love.

None of them won because they were able

to generate code faster than everyone

else. They won because they made sharp

product decisions under constraints.

They picked a narrow problem, executed

well, and said no to a thousand tempting

extensions. This is now all gone. That

constraint is disappearing and most

people are celebrating it as pure

upside. When the marginal cost of

shipping a feature drops close to zero,

discipline collapses. If a feature takes

10 minutes to generate and an hour to

patch, why not try it in five different

variations in just a few months?

Software products will replace

welloughtout concrete features with

half-baked ideas which will be a

nightmare to maintain. Don't be

confused. This will only increase the

need for software developers in the long

term. Ideas Oasmani was pointing out

that when the cost of generating

software drops, the surface area of

software expands. More teams build

internal tools, more departments

automate processes, and more experiments

get shipped. Entire industries that

previously relied on spreadsheets and

email threads suddenly have custom

dashboards and lightweight apps. If we

look at what happened in the past, we

learn that lowering friction will

actually multiply complexity instead of

removing it. The easier it is to create

something, the more things get created.

This is a classic economic principle

called the Jevans paradox. Where as a

resource becomes more efficient to

produce, we don't use less of it. We

find infinite new ways to use more of

it. Back to Dex's tweet, his next point

is actually really important. The

majority of workers have no reason to be

super motivated, and most of them want

to do their 9 to5 and get back to their

life. These guys will not use AI to be

10x more effective, even though

realistically this is not even possible.

What AI will do is help people turn out

their tasks with less energy spent,

which again can become a slippery slope.

The industry keeps framing AI as a

multiplier of ambition. In reality, for

many teams, it will become just an

energy saver. Coding, despite all the

mythology around it, is the easy part.

Writing syntax is mechanical. The hard

part is modeling the problem correctly,

defining boundaries, understanding data

flow, anticipating failure modes, and

designing systems that survive change.

When you write code yourself, you are

forced to simulate the system in your

head. You must understand why this

function exists, what assumptions it

makes, what state it mutates, what

happens when inputs are invalid. That

mental simulation is the training. If

that layer gets outsourced, the training

disappears. You still ship features and

you still close tickets, but you no

longer exercise the muscles required to

reason about complexity. So the two

people on your team who are actually

doing a lot of the heavy lifting already

are now also in charge of the slop code

everyone is producing. What's worth

noting is that even when you produce

code faster, you are still bottlenecked

by bureaucracy and a dozen other

realities of shipping something real. As

I mentioned in a previous video, the app

stores are filled with millions of apps

nobody uses. User attention,

distribution, discovery, and trust are

the real bottlenecks. None of those are

solved by generating another thousand

lines of code in 30 seconds. you'll just

have more dead apps in the stores. And

this brings us to the awesome trivia of

the day.

>> There is this theory of the dead

internet. I'll actually dive deeper in a

future video. In short, the dead

internet theory is a relatively recent

narrative claiming that most of the

internet is no longer driven by humans,

but by bots, automated content farms, AI

systems, and coordinated influence

operations. The core claim is simple.

Sometime around the mid 2010, the

organic human-driven web started to

decline. And what replaced it is a

synthetic layer of algorithmically

generated content designed to simulate

activity, shape opinion, and monetize

attention. Platforms like Facebook,

Instagram, Twitter, and Reddit have

documented bot networks, spam campaigns,

coordinated political influence

operations or engagement farming. And

just a few weeks ago, we saw the entire

AI community go nuts because of an agent

collaboration platform which turned out

to be a security nightmare with slop

content. Back to the video, the

conclusion of the video is pretty funny

because at this rate, the management

will end up wondering why each engineer

now costs $2,000 extra per month in LLM

bills. The truth is that the impact AI

will ultimately have on software

development is still unclear, but some

of the hype is already starting to fade.

Just a few days ago, IBM announced it is

tripling its entry-level hiring,

including software developers. Even as

many companies plan to replace junior

roles with AI, IBM's HR chief argues

that eliminating junior roles may

improve short-term efficiency, but

creates long-term talent gaps. Companies

that stop hiring early career workers

risk shortages of future mid-level

managers, higher costs from poaching

external talent, slower on boarding, and

weaker cultural integration. So IBM

believes companies that continue

investing in early talent now will

outperform the competition in 3 to 5

years. Sam Alolman also recently

recognized that a significant portion of

AI layoffs aren't due to the technology

taking jobs, but are instead part of an

AI washing trend where it's easier to

blame AI than to recognize weak

financial performance or passed over

hiring. This might be the case with

Block, which just laid off 4,000

employees. They named AI as the main

reason for their decision. But a lot of

voices mention revenue pressure and

financial issues as the real driver

behind the move. If you've made it this

far, I have a confession to make. Trust

me, I know you are all sick and tired of

the neverending AI news cycle. I share

the same frustration as many of you

guys. So, please let me know in the

comments if you find other interesting

tech news worth discussing. If you like

this video, you should consider joining

our community where I'm posting more

dedicated weekly content. Please don't

forget to smash all the buttons.