How Will AI Impact Business?

Introduction

AI’s impact has been and will continue to be vast. In the second article of this series (part one here), I’ll focus on the economic aspect of AI’s coming impact and save the ethical, geopolitical, and social impact for another time.

The original Henry Ford Moment

We are experiencing a Henry Ford Moment.

Let me dig into that statement. Our entire economic structure today — globalization, access to cheap products, American superpower, economic abundance, and consistent capital growth are all built on the back of Henry Ford’s genius to harness repetitive processes and machines into manufacturing automation.

While there were many manufacturers making cars before Henry Ford, each car was bespoke. A particular screw was machined, tweaked, and polished to fit a particular screw hole and no other. It took specialized labor to build each car, making their construction time consuming and expensive.

Instead of bespoke vehicles built by craftsmen, Ford focused on repeatable processes, consistency of parts, and speed of assembly. His famous assembly line was a means to an end. Repeatability and automation were his ultimate targets. In 1908 when the first Model T rolled off the factory floor, 514 minutes passed before the next car came off the line. By 1913, the last year of the Model T, a new car rolled off the line every 2.3 minutes. Ford not only changed car production but the production of all products everywhere.

Ford’s innovation was world changing; the resulting industrial might of the US’s manufacturing base meant the Allies won World War II, humans walked on the moon, the economy was globalized, and billions of people were pulled out of poverty.

I can’t emphasize enough the importance of this historical moment. In 1908, Ford set us on the path to leverage physical machines to their full economic potential, and in doing so, he changed our civilization.

The new Henry Ford Moment

AI is a revolution harnessing computing power and machine intelligence to their full economic potential. This is a revolution of a scale we’ve never seen before.

Let me explain.

The top companies in the world are those which provide the software or the products to run, make, distribute, or manage software products. Entertainment, knowledge work, communication, manufacturing, defense, transportation, services, and all other industries are dominated by the players that make the best use of information technology.

The top 5 public companies by market capitalization are Microsoft, Apple, Saudi Aramco, Nvidia, and Amazon. Four IT companies and a company that controls a huge portion of global energy production. Microsoft is a $3 Trillion dollar company. The company Henry Ford built, Ford Motor Company, isn’t even in the top 100.

The current economic hierarchy will not change. Instead, the gap between technology companies and everyone else will continue to widen.  The economies of scale for software are comically low compared to physical goods, and provide companies with the ability to distribute those products globally and instantaneously.

The AI revolution will take the economic fundamentals that make information technology products so lucrative and scale them up by hundreds of times, maybe more. And this revolution will happen at an exponentially reduced time scale.

It will dwarf the economic changes from mass-scale industrialization and reorder global economics and politics. Sundar Pichai, the CEO of Alphabet, believes AI will have a more profound impact on humanity than fire, the foundational technology of civilization itself.

He was not exaggerating. AI puts us on a civilization-altering trajectory. The pace of change will be absolutely shocking.

Immediate Impact

An immediately tangible example of this AI revolution is the ability to increase the economic output of software engineers, information workers, and other creative professionals. Software engineers in particular are generally obsessed with efficient workloads. “Never write the same code twice” is a mantra of any good engineer.  Those engineers can now leverage Large Language Models (LLMs) like Open AI’s GPT 4 to help them not write all of the code even once.

Creative professionals can ask an AI to make a logo using a design style, specific colors, and any number of other parameters, then fine-tune that logo. Researchers can parse huge volumes of text information into an LLM, fine-tune it, then ask it questions which it can answer immediately.   Film makers can leverage the newest video generators to create high fidelity storyboards today and soon full feature films.

The imagination and creativity from this group will be unleashed by AI. The companies that make the best and most productive AI models for these creatives, will realize exponential gains in market value.

This will also mean a rash of layoffs in some industries and heated labor battles in others. The long battle between Hollywood studios and the Writers Guild of America finally ended late last year with an agreement to put guard rails around the use of AI to replace writers. This was a prescient move by union leaders as they recognized the coming disruption to many filmmaking jobs.  The disruption of AI will still happen, but will now likely come from outside of the Studios.  With tools like Open AI’s Sora it might end up that the studios got the raw end of the deal.

The race to replace labor, empower labor, or create whole new types of work is here now.

Beyond the near term

If we consider an ever so slightly longer time horizon and pairing physical machines with AI, we’ll see another major increase in economic output.

Among all the uses of AI, the creation of humanoid robots to replace human labor will have the most significant economic impact. Large portions of economically valuable processes require some physical human labor: manufacturing a car, packing a box, mowing the lawn, cooking a meal, and building a house.  Robots haven’t replaced human labor in these activities because they were controlled by traditional software programs, which meant they learned too slowly and were unadaptable in dynamic environments.

In other words, they required an engineer to write, test, and implement new code to learn a new skill. Learning and thinking software paired with these machines will allow them to be adaptable at speed. It won’t take a round of programming for a robot to learn something new. It will only take someone showing it what and how to do it— whether that someone is a human… or another machine.

Emergent capabilities

Last and maybe most important, these AI models are showing emergent capabilities.  They are true intelligent things— machine intelligence, but real and working intelligence with a meaningful understanding of the physical and digital world. They are able to do work no one programmed, planned, or expected.

In a recent Harvard study researchers found an open source image generation model, Stable Diffusion, which had  been trained only using 2D images, created an emergent concept of 3D space in order to properly render images. Maybe more shocking was it seems to have also created an internal world-model in order to make life-like images of objects like cars, boats, or people.

Researchers are perfecting models which feed in multi-modal, or various kinds, of data into the training lifecycle.  These models, like Open AI’s Sora and Tesla’s FSD (Full Self Driving), are showing incredibly robust world-models which are able to simulate physics, 3D space, time sequence prediction, objects, and spatial permanence.

To be clear, these capabilities are not specifically trained, expected, or planned.  They are emergent and often unexpected.  No one on Earth knows exactly how these capabilities emerge, what is happening inside the neural networks, why scale dramatically affects capabilities, or if scaling will stop showing exponential improvements.

AI’s headline grabbing features are not due to engineering they are due to phenomena.

Conclusion

One societal note, which can’t be decoupled from the economic conversation, is the danger of concentrating economic value.  The nightmare scenario is for one or a few groups to control the vast majority of global economic value and production capacity offered by  AI and robotics.  The conversations necessary to avoid the consequences of such a nightmare have to start now, we’re simply too close to these technologies having massive economic disruption.

Competition in AI and robotics is critical to preventing economic overconcentration. The challenge is the enormous cost of the hardware necessary to train bleeding edge models and the undersaturation of talent.  Even top research Universities lack the resources to train new models at a meaningful scale, which limits their research to science projects using existing models.

This limit on innovation has consequences beyond economic concentration and stagnation.  Biases and beliefs of the researchers in the few companies able to build world class models are already bleeding into the training and fine tuning protocols.  As these models are used for more and more core computing and decision making across all sectors, the potential for concentration of power will be enormous.

These problems are billion dollar initiatives.  To solve them we need public-private partnerships at the State and Federal level to empower Universities and businesses of all sizes.

As an example, the forward thinking work by the State of Oklahoma to encourage AI development is remarkable.  The final report out of Governor Stitt’s task force is excellent and provides a model for other States.   The seeds of innovation will come from empowering more competition and more talent development.

Avoiding the economic pitfall of over concentration can bring a hopeful future.  While there will be major economic disruption across all fields; productivity gains, creative empowerment, and labor improvements can better all humanity.

Originally posted by Shane Kempton via LinkedIn.

Adapt to AI with Solutions from Phase 2

Our comprehensive strategies do more than leverage the creativity, potential, and innovation of AI; they incorporate the people, professionalism, and personal touch that matter to you, with elegant solutions that make sense and produce successful, sustainable outcomes.

Adapting to AI appropriately means embracing the right approach from the get-go. No matter where your business is in the adoption, implementation, or application of AI, the award-winning experts at Phase 2 can help develop and deploy AI tools that fit your unique needs. Contact us today to find out more about how we can integrate AI into your business.