Generative AI now has the potential to exponentially grow the quantity of content created. Articles, ads, images, videos, voice, graphs, 3D objects, music, scientific papers… all can be created today with AI. In a recent WSJ article, Nina Schick, author of a book on Deep Fakes said that “by 2025 or 2030, 90% of the content on the internet will be auto-generated”.
OpenAI, Midjourney and Stable Diffusion, are just some of the core technologies that are advancing the generative AI space, but the list of ‘application layer’ companies keeps on growing rapidly.
For example: TheGist, an Israeli startup launched this week with a $7M seed round, uses AI to summarise Slack channels and threads. GalacticaAI is a large language model trained on scientific papers. Type a text and Galactica will generate a paper with relevant references, formulas, and everything. Chula, is an AI assistant that creates graphics for presentations or Tome, which uses generative AI to generate presentation slides.
As I mentioned in my previous post on VC Cafe, “Generative AI will really take off when it moves from Playful to Useful“, I believe that we will see generative AI tools for every function in the company, and for every industry down the road. These tools won’t replace people’s jobs, but rather enhance our work, in the same way that we use Grammarly today to write better emails.
The Artificial Creativity Landscape by Anne Marie LeCunf tracks new companies building in the generative AI space and currently lists about 180 projects. Wired called a new Silicon Valley Gold Rush. Click on ‘source’ below to download a clickable PDF.
Another landscape of the growing number of generative AI startups can be seen in the following map and research by Base10 Ventures.
Generative AI is being adopted by developers and larger companies alike
It’s enough to look at Stable Diffusion SDK adoption on GitHub to get a sense of how excited developers are about this technology. In October 27, Stable Diffusion had more than 1 million downloads on Hugging Face.
In this graph you can see how steep has the adoption been, compared to other technology trends.
But it’s not just startups and developers that are adopting these new ‘creative automation’ capabilities. Larger companies are also pouring resources in this space. Let’s look at a few examples:
- Notion AI – Notion now offers users ways to create content faster using generative AI
- Canva launched an AI image generator as part of their online design suite
- Microsoft added Dall-e 2 to their office suite with their Designer app and Image Creator
- Github – added copilot to generate code with AI, and more coming in Github Next
- Shutterstock is using Dall-e 2 and Getty partnered with Bria to help users generate stock photography from text prompts
Many more applications are likely on their way. It’s interesting to think what the most popular consumer and enterprise products could look like if they added generative AI capabilities. Wikipedia, Spotify, Shopify, Instagram… but also for the world of work: Google Docs, Asana/ Monday, Wix, etc. It feels like the technology has gotten good enough and I believe that it’s a matter of time until this happens.
You can get a sense of the magnitude of impact of generative AI text/ language can have on various sectors and jobs, in this Forbes piece:
The ability to automate language thus offers entirely unprecedented opportunities for value creation. Compared to text-to-image AI, whose impacts will be felt most keenly in select industries, AI-generated language will transform the way that every company in every sector in the world works….
Massive disruption, vast value creation, painful job dislocation and many new multi-billion-dollar AI-first companies are around the corner.
Rob Toews, “The Biggest Opportunity In Generative AI Is Language, Not Images“, Forbes
Will creation replace search?
Remember the early days of the web? I used to learn about new websites from recommendations, or magazines/ newspapers that would feature a few websites at a time. Then came the directories and as the number of websites started growing exponentially, it gave a rise to search engines. I remember using 5 or 6 of them, before Google, the best one of them all came along (Excite, Webcrawler, Altavista… rings a bell?)
Going back to where we started, as these Generative AI tools become more widely distributed, the effort required to create content will go down, and the quantity of content will significantly go up. What does it mean for search? Are we going to continue to use Google to find all this new content in real time? Or will it be faster to just create the answer to our question? Will we search for websites or prompts? and when will we need a human in the loop to make sure the content generated isn’t junk?
Given the dominance of Google in the search space, it didn’t change much over the past 20 years. A few tried and failed, and those that still compete with Google on search (like Bing), have nearly no market share. I spent many years in my career as a product manger, specialised on search, and that’s why I get easily excited when I see a new innovative approach. Take a look at Metaphor:
Metaphor is a search engine that understands language – in the form of prompts – so you can type what you’re looking for in all the expressive and creative ways you can think of.
The model that powers Metaphor is trained using a form of self-supervised learning, the same paradigm behind models like Stable Diffusion and GPT-3. Stable Diffusion tries to generate images based on their captions, GPT-3 tries to predict the next word based on the previous ones, and the model behind Metaphor tries to predict the next link on a webpage based on all the words that come before it.
I’ve been playing around with it, and really like the ability to find ‘related URLs’ given a URL prompt, or the natural language recommendations. For example, try searching for “My two favorite blogs are VC Cafe and”
Considerations for investing in Generative AI
Generative AI startups have attracted hundreds of millions in investments. However, it is still early days for the technology and it will take time for it to reach its full potential. From an investor perspective, my concerns investing in the generative AI space are mainly around:
- Copyright – upcoming litigation against Github’s copilot over the use of open source code for training or the RIIA (The Recording Industry Association of America) suing music generative AI tools for copying infringement.
- Ethics – putting the risks of job loss and automation aside, Stability AI has already been used for the creation of AI generated porn and deep fakes. Lack of regulation and potential bad actors create a big liability in this space.
- Dependency – the dependance on someone else’s API to build the core product of a startup (say GPT-3 or Dall-E 2) can put the startup at risk. What if the API changes its T&Cs or massively increase prices? That’s why I’m excited about the rise of open source
- Defensibility – If competitors are mostly building products on top of the same APIs, the product with the best user experience will win. It’s hard to create significant defensibility in generative AI, and therefore it’s safe to assume a lot of competition for each category. Being fast to market, providing a top user experience and cracking a sustainable business model is key.
Shameless plug
At Remagine Ventures, our first investment in this space was in HourOne in 2019, followed by Munch last month. We believe that the time to start innovative companies in the Generative AI space is now. If you’re a founder, or team, looking to start something in this space, I’m happy to chat.
I'm a former general partner at google ventures, head of Google for Entrepreneurs in Europe and founding head of Campus London, Google's first physical hub for startups.
I'm also the founder of Techbikers, a non-profit bringing together the startup ecosystem on cycling challenges in support of Room to Read. Since inception in 2012 we've built 11 schools and 50 libraries in the developing world.
- Gaming Investments Level Up: Early Signs of a Resurgence - April 24, 2024
- What could go wrong with Vinod Khosla’s techno-optimistic vision of the future - April 22, 2024
- Keeping it Real: The Struggle for Objectivity in Tech Reviews - April 18, 2024
