AI: Where Are We? Where Are We Going?

LLMs collapse the cost of code

This is AI's first killer app

👤

Human

intent

AI

translates

code

⚙️

Systems

AI will become the new integration layer.

Our digital infrastructure will get rebuilt.

13

Andrej Karpathy

Former Director of AI at Tesla. Founding member of OpenAI.

December 2025

"I've never felt this much behind as a programmer. The profession is being dramatically refactored... I have a sense that I could be 10X more powerful if I just properly string together what has become available over the last ~year."

January 2026 — One month later

"I went from 80% manual coding to 80% agent coding. I'm mostly programming in English now... This is easily the biggest change to my coding workflow in ~2 decades. It happened over a few weeks."

"There are decades where nothing happens; and there are weeks where decades happen."

— Vladimir Lenin

14

Slowly, then suddenly

15

This is now the biggest CAPEX boom in a hundred years

Tech capex in 2025 exceeds the Manhattan Project, the Apollo Program, and the Interstate Highway System — combined.

This is not hype.

When capital moves at this scale, it reshapes industries.

16

Demand will naturally explode

First, we will see cognitive work shift from biological to silicon. This is a conversion of existing demand.

Then, as compute gets cheaper and new use cases get unlocked, Jevons' paradox kicks in — and demand for compute continues to boom.

17

For companies, this is still the beginning

CEO confidence in revenue outlook hits five-year low — as AI becomes a defining divide between leaders and laggards: PwC 2026 Global CEO Survey

January 19, 2026

Only three-in-ten (30%) CEOs are confident about revenue growth in 2026 as most struggle to turn AI investment into tangible returns.

One-in-eight (12%) CEOs say AI has delivered both cost and revenue benefits, while companies that have scaled AI with strong foundations are pulling ahead.

What impact did AI have in the last 12 months?

Cost

Decrease No change Increase

1%

13%

12%

0%

42%

8%

1%

12%

9%

Decrease No change Increase

Revenue

18

So this is where we are now

CEO chant meme: Who are we? CEOs. What do we want? AI. AI to do what? We don't know. When do we want it? Right now.

The challenge is now bridging the gap

19

Tasks get automated, not jobs

Adoption will happen in three phases:

1. AI helps you complete some tasks.

2. AI completes the tasks for you.

3. AI does tasks that you could not do before.

20

Phase 1

AI helps you complete some tasks

You drive. AI accelerates.

Writing, editing, and summarizing.
Code completion and review.
Data analysis and visualization.
Meeting transcription and action items.
Research assistance and translation

GitHub Copilot

ChatGPT

Claude

Microsoft Copilot

Cursor

Google Gemini

21

Phase 2

AI completes the tasks for you

AI drives. You supervise.

Autonomous customer support.
End-to-end code generation from specs.
Document processing and reconciliation.
Automated reporting and analytics.
Enterprise data integration and action

Salesforce

ServiceNow

Intercom Fin

Devin

Klarna AI

Harvey

Palantir

22

Phase 3

AI does tasks you could not do before

New possibilities arise.

Protein structure prediction.
Drug discovery at molecular level.
Autonomous driving.
New media generation (image, video).
Pattern recognition across impossible data scales

Isomorphic

Recursion

Waymo

Midjourney

Runway

Palantir

23

Skills for the agentic workplace

Prompt Thinking

Oversight & Judgment

Workflow Design

Data Literacy

Problem Decomposition

Domain Expertise

Adaptability

None of these are "how to code" or "how to build AI."

They're about how to work with AI. Adaptability is the meta-skill — because the tools keep changing.

24

The real challenges

This must be a C-suite priority

IT will freeze the ground around it if not driven from the top.
This is a change management issue as much as a technology issue.

Nobody knows anything (yet)

These are new tools that few know how to use well.
We're early in the game — mistakes are expected and valuable.

AI must come from within

Cannot be bought "out of the box."
Must unlock and leverage YOUR organization's unique tribal knowledge.

The tools change with scale

What works for an individual won't work for a team.
What works for a team won't work for the organization.

Organization

Team

Person

25

A glimpse of what's ahead

LLMs → VLMs → VLAs → Recursive Self-Improvement

LLMs: Read and write (GPT, Claude)

VLMs: See + understand

VLAs: See + understand + act

Next: Agents that improve themselves through reasoning

The pace of change is accelerating, not slowing.

26

"You're not going to lose your job to AI. You're going to lose your job to someone who uses AI."

Jensen Huang

CEO, Nvidia

27

What is AI?

"Computer" used to be a job

Tools are leverage

AI is leverage

It is a new type of intelligence

Human Intelligence

Deterministic Computing

AI / Stochastic Computing

AI is like Rainman

AI is also like Leonard in Memento

So What?

What does this new kind of intelligence
mean for organizations?

Why do companies exist?

Information technology pushes down coordination costs

As code went everywhere, it became a bottleneck

And then it wasn't