The Cambrian Explosion of Software 3.0
Generative A.I. and the story of Applied LLMs/Transformers
The Cambrian Explosion of Software 3.0
We may be at a unique turning point in the history of artificial intelligence and may not fully realize it yet.
How Generative A.I. will Usher in the Machine Economy
When I'm not writing for my lead A.I. Newsletter at A.I. Supremacy and my related Futurism coverage, I'm writing on LinkedIn (where A.I. Report has crossed 210,000 followers) and beehiive. It's important for me to diversify, as I try to make sense of new trends.
Let's assume for second that Generative A.I. continues to bloom with the announcement of GPT-4 in 2023.
From gaming to code generation, video to sales and marketing - what if A.I. really does disrupt how we not only create content but create companies, scale products and manage tools and do engineering in the 2020s?
What if Transformers and LLMs do find more than just utility, new ways of combining with compute and new products that make A.I. truly useful? As Venture Capital continues to sing the praises of Generative A.I, we have to assume that the realization of a Software 3.0 new paradigm may indeed be upon us.
Prediction: Funding is Going to Increase with the Utility of Generative A.I.
Even OpenAI is funding early Generative A.I. startups. Microsoft may again have first dibs on GPT-4 with increased funding for its "sponsorship" of OpenAI, once conceived as a non-profit A.I. lab. A.I. has a legendary way of making people greedy.
But who can blame them? Over $100 billion has been invested in AI startups since 2020, and funding doubled in 2021. There are 102 AI unicorns in the US and 38 in Asia.
With the "Generative A.I." hype now in full-force, there will be more funding for A.I. startups in 2023 and 2024 than many can expect or even imagine. So why might we say that there is indeed a Cambrian Explosion of Software 3.0 startups and momentum?
Base 10 VC sees a Generative AI universe has over 300 companies, each Venture Capital Fund is only able to see a fragment of the entire picture. a16z realize Generative A.I. will eventually extent into coding and the development of better games, faster and cheaper than ever before.
At a time when the Metaverse and crypto look vulnerable (AF), AI’s contributions to transportation, finance, and media already dwarf any value recognized from crypto, according to Professor Scott Galloway, you can read his Op-ed on A.I. which I enjoyed very much this past weekend. There's no mercy or malice when it comes to creating new paradigms, Silicon Valley needs them to prop itself up.
Take any segment of Generative A.I. Software 3.0 startups and you will see twice as many by the end of 2023. Take any imaginable utility of Generative A.I. in 2022, and you will find dozens of new use-cases in 2024. This may be an exponential increase in A.I.'s fundamental utility we'll experience in the 2020s, a result of a huge increase in R&D, academic papers and momentum in research collaboratives and global collaboration in academics.
You cannot even visualize what Software 3.0 will become in this Cambrian Explosion of use cases.
Source: a16z Games
Just in the past few months we’ve seen progress in AI capabilities that has left even skeptics impressed. A “golden decade” one researcher called it. Generative A.I. is the name that best shows the exponential release of what A.I. can finally do in the field and promises to do soon.
LLMs and Transformers will be doing a lot more than lending Authors a helping hand in the years ahead. Founders are deploying complex foundational models against real-world problems that are creating new software applications and fundamentally changing the way old industries work. But if 2022 gave us the imagery of what this might look like with Stability.AI, Midjourney, DALL-E 2 and others, 2023 will show us the range of its utility.
We may be witnessing something rather special, a significant moment that only comes around maybe every 13 years in the world of technology. A moment that builds upon new foundations, a re-set of what is possible in the Cloud, in software and with A.I. We are a society increasingly embedding machine learning into how we work, live and create products and build businesses.
Since Generative A.I. will improve productivity and utility, the highs will be higher and the lows won't be as disappointing as bets like Robo-taxis or VR worlds. Generative AI investment multiples suggest we are at peak Gartner hype. While you might actually expect the “trough of disillusionment” to be shallower than usual, that means the plateau of productivity is quickly approaching for AI and generative AI, in particular; we can all expect improvements to be exponential and power a transformative platform shift.
One of the earliest uses of generative AI has been as a programmer’s aid. The way it works is that a model is trained on a large corpus of code (e.g. all the public repos in GitHub) and then makes a suggestion to a programmer as they code. The results are outstanding according to Microsoft, lord and chiefs over GitHub and OpenAI itself with promises of more funding. OpenAI may as well be to to Microsoft what DeepMind is to Google, a leading part of how they integrate A.I. and do cutting-edge research. Yet Microsoft's ability to integrate A.I. utility into its products is infinitely more diversified including in the future of coding and gaming itself. This means that Microsoft doesn't just have the best dividend in BigTech, it's the most likely company to be synonymous with the Generative A.I. trend.
It's hard to overstate how unfair Microsoft's advantage is in leveraging Software 3.0 with its already developed low/no code apps and platform. Forget NFTs and Bitcoin's volatility, think of how the Cloud and software is evolving in the 2020s. Artificial intelligence is widely believed to be the transformative agent, and if it is, we may be at the beginning midst of a great leap forward in AI, and that this tech will be transformative, not just lucrative, thanks to its utility. Microsoft with its subscriptions and Azure Cloud focus, knows a thing or two about offering businesses utility.
Generative A.I. Early Adopters Emerge in 2023
In 2023, while the world is in a global recession, A.I. and automation will prove it never sleeps. Jasper.AI, Stability.AI., Microsoft and many other A.I. Generative native companies will come to the foreground. Canva and Nvidia will invest in the technology, so will countless others.
This year’s State of AI report shows a hundred slides of accelerating progress. In 2020 there wasn’t a single drug in clinical trials that had been developed using an AI-first approach. Today there are 18. Drug development, genomics, biotech and computational biology will all feel the impact of Generative A.I. over time.
AI has been in pop culture for generations, but it's role will go from promise to utility, in as if a blink of an eye. Stable Diffusion is just the beginning, text-to-image a showy foretelling of what becomes now possible. Automation and robotics takes a leap, even as our computer and LLMs converge on something truly special.
The way we think of the future will begin to change. The hybrid-human AI workforce comes into shape. Even how developers manipulate code will be re-made.
The Ubiquitous Self-Fulling Prophecy of A.I.
There comes a point where A.I. own acceleration underpins all that we do. A point of ubiquitous no-return. The way we think of the future will begin to change. The hybrid-human AI workforce comes into shape. Even how developers manipulate code will be re-made. Just as the Cloud matures and software itself evolves. There is a self-fulling prophecy in A.I. hype of the century, the 21st century. And truth be told, we do not know exactly what kind of a future we have in store for us.
So is the hype justified? Just as compute power channels new possibilities so do more efficient and larger LLMs. The generative AI opportunity is worth trillions of dollars. Sequoia estimates that a 10% improvement in knowledge work productivity thanks to generative AI is worth trillions of dollars in value. Generative AI is going to touch the efforts of billions of knowledge workers, including paralegals, lawyers, bookkeepers, accountants, software developers, marketers, and consultants. The opportunity presented by the technology to streamline writing, design, research, summarization, and coding is monumental.
Those companies that achieve a state of A.I. Supremacy in generative-A.I. will have significant competitive advantages over their rivals. If you thought Microsoft's venture into the cloud with Azure was impressive, just wait until you see their assault on Software 3.0 with Generative A.I. and just how much revenue their sponsorship of OpenAI with that first $1 Billion dollars is really worth.
The pressure of China vs. the U.S. will actually speed up the acceleration already embodied in the utility that Generative A.I. will bring to the evolution of the next layer of the internet. An internet that's more dynamic, transformative and of embodied machine learning and human levels of creativity.
Even for engineers, data scientists and software developers it's hard to fully fathom what Generative A.I. will do to the future of coding. Within a generation, it could be significantly different.
The way businesses use A.I. is changing in the era of Generative A.I., low-/no code, RPA, automation and MLops. We are entering a new era of artificial intelligence at scale, accelerating faster in utility than was possible before Transformers came into being.
A transformer model is a neural network that learns context and thus meaning by tracking relationships in sequential data like the words in this sentence.
First described in a 2017 paper from Google, transformers are among the newest and one of the most powerful classes of models invented to date. They’re driving a wave of advances in machine learning some have dubbed transformer AI. Eventually this evolved into the Generative A.I that we now called today the Software 3.0. movement.
Within a few months of one another, three different image-generating AIs were made public. Dall-E, Midjourney, and Stable Diffusion are variations on a theme. You enter text, and in a few seconds the system generates an image.This has given us new capabilities within a framework of human creativity. Many expect Generative A.I. to give us new powers in multiple industries more or less simultaneously in the 2023 to 2026 period as more Generative A.I. native startups are born. They will need some time to mature, but their capabilities may bring a new era of utility to what has been a nearly sci-fi realm of what we hoped one day A.I. would become.
The Accelerating Arrival of the Machine Economy
Since killer apps for generative AI will be domain-specific, just as DALL-E, Stability A.I. and Midjourney changed how we view images, so too will the leading startups change how we do video, search, code, music, sales, gaming, tools, marketing, design, product and so forth.
As society looks for ways to improve productivity it will turn to automation, robots and Generative A.I. The startups and companies that offer solutions to some of society's most pressing concerns like maintaining GDP growth will be native to the Machine economy, a world where humanity is more embedded with A.I. and data than we are today. This transition usually takes many decades but Generative A.I. could speed it up considerably.
Humans are good at analyzing and building things. Machines will become even better.
While GPT-3 was all about foundational models and possibilities, GPT-4 will bring more tangible drive to the utility of what Generative A.I. can become across industries. Existing companies will leverage GPT-4 and the startups that utilized ever refind and efficient Transformer architectures and LLMs to bring new scale to compute and human-AI productivity.
I believe GPT-4 will be announced in the first half of 2023 by OpenAI and many other A.I. labs are just months behind it.
But humans are not only good at analyzing things and building things —we are also good at creating.
We'll become better at doing everything ultimately with A.I. This is when the Machine Economy proper is truly born. Software 3.0 enables it to come into being. Like a fabled world of Quantum advantage or the Singularity, the Machine Economy is an A.I. native era of human activity. Machine Intelligence begins to optimize each industry, every profession, ever tasks humans do or will decide to do, impacting every project we will decide to undertake.
Up until recently, machines had no chance of competing with humans at creative work—they were relegated to analysis and rote cognitive labor. Narrow A.I. felt relegated to the most tedious tasks and we believed only repetitive human laborers were vulnerable to disruption. How wrong we were!
I may resist the urge to predict exactly how generative AI will impact the future industries However, history suggests that new tools tend to expand rather than contract the definition of what is possible. In the largest sense I believe Generative A.I. is an ultimate ally to the essential human creativity and thirst for progress that best defines our spirit. The human-machine empowerment whether you agree with it or not aesthetically, morally or philosophically, is the end-game of the technological society.
Our cognitive and sociological evolution is now intrinsically tied to the evolution of our artificial intelligence itself, our one creation that may help define best our future as a species and a technological civilization that may or may deserve to live a few centuries or millennia longer.
Generative AI’s applications touch a variety of end users and over the course of the following years they will touch a greater number of people who use the internet and a greater number of businesses and industries. There is a broad based consensus among analysts, venture capitalists and academics that this trend can grow as foundational models become more efficient and as LLMs become even bigger and more efficient.
What will then be the impact of this machine economy on the world? Will we live in a more automated AI-software based digital existence? The Machine Economy implies that digital transformation will mean people are augmented by A.I. to a greater degree than before. In the Machine Economy digital-first organizations become AI-first companies. Generative A.I. is the catalyst to a Machine (learning) driven economy.
The Software 3.0 revolution is thus already underway. And by 2030, AI will contribute an estimated $15.7 trillion to the global economy, more than the current output of China and India combined. Even our conception of what automation and a data-first world looks like will shift with the impact of Generative A.I. For some industries it's not yet clear what that will look like. People will work differently and A.I. will make us more productive. New software tools will emerge that are no/low code that lowers the bar to adoption.
We have to note that these systems could supplement or replace human creators in many sectors. How content is generated whether it be images, writing, video, code and so forth whether for blogs, marketing, sales, art, ads and other uses - could be greatly facilitated by generative A.I. tools that will only get better for here on in. This is a shift in how we view A.I. in society, business and the future.
Generative AI is well on the way to becoming not just faster and cheaper, but better in some cases than what humans create by hand. Every industry that requires humans to create original work—from social media to gaming, advertising to architecture, coding to graphic design, product design to law, marketing to sales—is up for reinvention.
To what degree the Machine Economy replaces the current system or to what level of sophistication and how quickly it attains Generative Advantage (equal or better to humans) remains to be seen per task and per industry. Pragmatically, Sequoia notes that certain functions may be completely replaced by generative AI, while others are more likely to thrive from a tight iterative creative cycle between human and machine—but generative AI should unlock better, faster and cheaper creation across a wide range of end markets.
This is what the Machine Economy is and it's expected Generative A.I. and software 3.0 will indeed build a new normal for how people and knowledge workers accomplish many tasks. We can say that the dream of Venture Capitalists with Generative A.I. is clear, the dream is that generative AI brings the marginal cost of creation and knowledge work down towards zero, generating vast labor productivity and economic value—and commensurate market cap.
Though there will be real pros and cons as with the invention of any new tool. The past decade has seen tremendous progress in the field of artificial intelligence thanks to the resurgence of neural networks through deep learning. Generative A.I. is ushering in an age of A.I. utility that frankly, we've never truly seen before.
Artificial neural networks, which draw inspiration from real biological neural networks, seemed like a promising approach for much of this time, but ultimately fell out of favor in the 1990s. Since the advent of Transformers around 2017, LLMs have show increasing abilities to mimic human level performance.
Functionally, as the models get bigger and bigger, they begin to deliver human-level, and then superhuman results. Between 2015 and 2020, the compute used to train these models increases by 6 orders of magnitude and their results surpass human performance benchmarks in handwriting, speech and image recognition, reading comprehension and language understanding. In 2023 something interesting will happen, GPT-4 will give us yet a new benchmark of what is possible.
2022 brought us a lot of change actually in a single year, it compressed what would have taken years previously. Over the course of just a few months compute got cheaper. New techniques, like diffusion models, shrunk down the costs required to train and run inference. The research community continued (continues) to develop better algorithms and larger models. Developer access expands from closed beta to open beta, or in some cases, open source.
The Generative A.I. story is one for the future. But that future is quickly becoming our present. For developers who had been starved of access to LLMs, the floodgates are now open for exploration and application development. Applications begin to bloom. We are fundamentally living in a world where A.I. is increasingly more accessible and democratized. This is the beginning of the Machine Economy proper and the wave of software 3.0.
How to Predict the future of Generative A.I?
For us in 2022 to imagine what will be possible in 2025 and 2030 is likely fairly difficult. It depends on things like GPT-4 and GPT-5 and how many more startups take up the challenge. My guess is that things will move slightly faster than expected as R&D, open global A.I. research communities and A.I. startups all proliferate trying to capitalize on the window of commercial opportunity.
I'm not sure if Venture Capitalists make the best futurists as they have financial incentives to spin a good tale. But as someone without the ability to have financial incentives about the success of A.I. as a whole and new kinds of startups, I believe Generative A.I. is likely what manifests the Machine Economy of the mid 2030s. It is in some respects, over the course of the next fifteen years, a fundamentally different world. Transformers will turn out to be a disruptive technology at scale and in mass adoption.
Thanks for reading. Follow me on LinkedIn. You can subscribe to A.I. Supremacy, if you want to support my work and gain access to premium regular content.