This web app uses cookies to compile statistic information of our users visits. By continuing to browse the site you are agreeing to our use of cookies. If you wish you may change your preference or read about cookies

December 22, 2022

AI as a commodity in 2022: How did it change our world?

Hands of God and AI (Polaroid retro style) – Author: Pedro Trillo & DALL-E

There is no doubt we are living in times of extraordinary change, 2022 will end soon, and it will be the year in which Artificial Intelligence became a “commodity” at zero or close to zero cost, the same year in which gas and electricity (the real commodities) reached the price of gold, the world is upside down, do not look for an explanation, we have entered a random and exponential cycle, in which anything can happen.

Embrace uncertainty, and make friends with it; we have no choice; I’ll tell you how I do it; my thought is that everything is uncertainty, and I relate to it in a very natural way; I have integrated it into my consciousness to feel comfortable with it when I rarely find a minimum period of certainty, days or weeks, is when I worry, the exceptionality confirms the rule.

I think it is much more likely that we will soon reach the singularity with a possible GPT-X than that we humans will agree on how to manage the pressing energy demand we need; moreover, we will ask that AI How do we solve the problem of the earth’s energy? And it will answer us, something that will be so obvious; we will say, how have we not seen it before?

It has been a crazy year for those of us who are dedicated to or have a Generative AI company; it has been literally impossible to follow everything that has happened this year; from May onwards, we went into a full-blast mode, and practically every two weeks a new model has been released, a new API has been released, or barriers and walls that seemed impossible years ago have been crossed.

We are in a profound change, and we are mutating from the Information Age to the Imagination Age.

From text-to-x:

  • text-to-text
  • text-to-code
  • text-to-image
  • text-to-audio
  • text-a-video
  • text-to-music
  • text-to-motion
  • text-to-gif
  • text-to-3D
  • text-to-NFT

The models listed below are mostly models that have emerged this year or have been in the pipeline since 2020 and have released powerful new versions this year:

  • AI21
  • mdm
  • gpt-J
  • gpt-3
  • x-clip
  • bloom
  • cohere
  • gopher
  • dall-e 2
  • craiyon
  • tabnine
  • jukebox
  • anthropic
  • codegeex
  • gpt-neox
  • nvidia get3d
  • dreamfusion
  • stable diffusion
  • meta make-a-video, opt

Regarding OpenAI, the list keeps growing, and we are reaching extraordinary levels of development:

  • gpt-2
  • gpt-3
  • dall-e
  • jukebox
  • whisper
  • openAI CLIP
  • openAI codex
  • chatGPT

Stability AI appeared this year as its open-source option with Stable Diffusion, and this was when the market became dynamic; those who follow the blog know that I am a big fan of open source, they are the ones who have exploded the market, and have caused drastic changes in the industry, so there must be the open source option.

Clicking on the image below, you will open a video summary of everything that has happened until October 2022, 2 months left, which have been very intense. Still, this content of Marten Kuipers of DEPT illustrates and summarizes, in a very graphic way, all the milestones that have been achieved.

The rise of AI – Marten Kuipers from DEPT

To complement all the 2022 summary content below, I show you the state-of-the-art industry report, which does not include the year-end surprise with ChatGPT:

State of AI Report 2022

I have to close the 2023 roadmap, and I can’t; it’s impossible; I let myself go; I keep a fluid communication with the OpenAI team, who don’t let go either; I can only make plans for a month ahead; with a very high probability of change, it’s impossible to keep up, we thought that GPT-4 would be the next thing for January 2023. Still, no, ChatGPT appeared on November 30, a brutally good model that fine-tuned all the data coming from GPT-3; it would be a mistake to call it AGI (Artificial General Intelligence).

But a more significant mistake would be to underestimate the power of this tool in general use; in 4 days, they got 1 million users, who literally knocked down the servers. To this day, ChatGPT marks a milestone in which the entry barrier to these magical technologies has been wholly lowered; anyone can access ChatGPT, a model of the open and accessible moment, which has propelled the normalized use of AI by millions of Internet users.

The changes are coming, not when four tech geeks build with these tools, but when your brother-in-law sends you a picture this Christmas of “A Picasso-style donkey with a Santa Claus hat flying through the Milky Way.”

OpenAI needs millions of validations for this model, they use RLHF (Reinforcement Learning from Human Feedback), and this is the reason for its opening to the general public, which has caused an unprecedented viral in social networks naturally and organically; I have not seen any paid advertisement from these companies.

Then we will have to run again in 2023 to make the new integration when they release and officialize the API, so we go all year, running after a greyhound.

The absolute state of AI research rn tweet de Andrew Carr

We are facing a historical milestone, as was the emergence of the Internet, social networks, the iPhone, or Bitcoin; we are mutating our use of the Internet from GOOGLE, with easy questions, complex results, TO CHATGPT, with a complex question, easy result. This is the first time in 24 years of Google’s history that we are all seeing the first disruption coming to them in 2023; it’s not that Google doesn’t have good AI, nor have the ability to pull off their ChatGPT, let’s not kid ourselves, it’s that OpenAI (Microsoft), is doing exceptionally well, in time management, for Google generative AI hits the waterline of their business model based on links and advertising.

Then we are going to witness 2023 with a plummeting of social networks like Facebook or Twitter, a Google that will have to act very fast, and a new technology that will burst next year in an explosion of change, which should be in all product roadmaps of 96% of current software platforms, to understand, the proposal is so disruptive, that any software or platform that is not integrated with these models, is obsolete.

If you think about it, we have been working for 20 years with software that are actually dumb boxes to be filled by humans, edited, shared, and little else. The challenge for the technology industry in 2023 will be to transform dumb boxes into bright boxes, and in this bypass, there will be winners and losers.

As I write this article, I have JIRA open in my browser; JIRA is a helpful software for software product management; you write the requirement you want, you plan it, you assign it, and you share it; well, that blank box you see in the image, will have a button that says generate text with AI, and so thousands of software and apps that we use in our day to day, we all follow the same pattern.

You will find ChatGPT and the substitutes in your operating system (it will not be like Cortana or Siri). Still, you will interact with it by voice; you will ask it, send me an email, make me an offer document, create a new functionality in JIRA, you will find it on the Internet as a search tool, you will find it in any software that is based on text, image, sound, and video, the four fundamental pillars.

You will interact with general-purpose AIs like ChatGPT and with AIs specialized in particular topics like Vizologi; it may happen that we will reach a level of AGI development so good that we will all go to the beach and leave a bot simulating our voice and talking to our Windows 15.

The summary idea is that we will end up with a super-app, a Chinese AI Wechat that practically assists you and helps you with 96% of your daily tasks; the problem is that, that we will reach an AGI that adds so many services and functionalities that it will somehow leave the whole technological market in check.

A fundamental characteristic of generative AI software, whether general or specialized, is that we do many little things at once; we are aggregators in the way of several services that in previous software versions were solved 1 to 1, one software for one service, with Generative, AI we aggregate services.

Then 2023, full of surprises, awaits us in the technology industry; we will see movements, new players, acquisitions, alliances, closures, and major disruptions. Designer, a new Microsoft application for the creation of self-generative designs, will eat Canva’s market share, a very consolidated startup for years; Adobe, image banks like Shutterstock, and all the big players in the industry will be redrawing their AI strategy for 2023.

Whoever acts late will be out of the market, which is going at the speed of thunder. The first disruption will come to the major software players in 2000; then, it will come to the rest of the knowledge-based industries.

Those of you who follow the blog regularly know that I told you more than a year ago that we were no more than 30 startups worldwide, the ones that opened the generative AI market; there was Jasper, copy.ai or Vizologi, among others, the first two with many millions behind the scene, all said, that’s why they go as they go, today the number of new generative startups has multiplied x10. Next year it will be multiplied by x100.

Sequoia – Generative AI

Every week a new generative AI startup is born, and they are coming out like mushrooms at a particular level; I have to tell you, it has been a good year, and the end-user has evolved a lot in understanding and learning with technology, the root came by the models of image creation, being the very visual experience, it helped us a lot to text startups to evolve and make a leap in the understanding of our products.

In the VC world this year, the vertical was born, and the anecdote is that it was not known what name to put; since 2016 in this blog, I have always referred to it as Generative AI, and no one has complained, but there was indeed no formalized or common name. Still, finally, Sequoia has baptized it as Generative AI or Generative Tech, as the name of the new vertical that was born this year. In fact, we even met with Sequoia (San Francisco) in a global call they made about generative-based startups a year ago; this was impossible.

Let’s understand that we are living in a world where generative AI is solving very complex problems like mp3 audio compression; Meta released two months ago a codec that generated an AI that compresses audio x10 without loss of quality, that is a problem that took more than ten years to expert researchers, it took minutes for the AI to solve it, offering a solution ten times more powerful.

Those of you who have seen this year’s Netflix series, The Playlist, tells the story of how Spotify was born; the trigger was that technological development allowed audio to be compressed to an mp3 format for subsequent encapsulation and distribution via streaming over the Internet, the period between 2000 and 2005, Napster, Emule, Pirate Bay, which distributed free music, were born, legal problems and lawsuits multiplied, to the point of putting the recording industry in check, then Spotify appeared with a business model that a priori convinced all parties.

Well, Generative AI makes this historical fact anecdotal; we will see great struggles and wars during 2023, copyright, copyrights, etc.; this fact tells us that a massive change is approaching, something very transformative as was once the birth of streaming audio and video, the impact of what is coming is tremendously superior to what happened in those years, once it was the audiovisual industry, now it is all companies and professionals based on the information society and knowledge.

The north star of generative AI is that it reduces to zero the cost of knowledge-based labor, generating enormous labor productivity along with extraordinary economic value; the real abundance will come this way; what we have been living has not been abundance; it has been a mirage based on an accumulation of infinite debt.

If at this point of the article, you think that all this is hype, a bullshit generator, that a crack pattern similar to the Metaverse or Cryptocurrencies will happen, and that the “fashion” of Generative AI will fall sooner or later, I refer you to the article of Cacerolas, postcapitalism and pouting tales; Generative AI is here to stay, it is not hype, it is PRODUCTIVITY with capital letters when you hesitate to bet on one technological trend or another, think about the assumption I raised in that article:

“The maxim that underpins the system in which we live focuses on producing more in the shortest possible time, reducing marginal costs, and increasing the profit margin.

On these four legs of the table, and following these premises to the letter, we have arrived at a global production model in a worldwide framework and with a technological level that exceeds and changes the very nature of the system.”

The technologies that are imposed are those that save time, reduce costs, and increase profits; there is no more fabric to cut, let’s not complicate our lives; six years after writing this article, I would add that triumphs are imposed; naturally, those technologies that are EASY, SIMPLE, AND SIMPLE, metaverses and cryptocurrencies are DIFFICULT, COMPLICATED AND ARTIFICIAL products that the big market, the millions of Internet users on the street, are not able to assimilate.

As every year, I take the opportunity to wish you happy holidays; with a retrospective look, I say goodbye to the blog until next year, closing 2022 that has been incredibly prolific, as well as millimetrically accurate in all predictions, both personally and professionally the blog has given me a lot this year, I hope it has helped you too.

Thanks for reading me for one more year!

Pedro

Share:
FacebookTwitterLinkedInPinterest