Read on Twitter
Five Advancements Powering the Future of Content Creation
A thread
https://www.vccafe.com/2021/02/02/five-advancements-powering-the-future-of-content-creation/
A thread
https://www.vccafe.com/2021/02/02/five-advancements-powering-the-future-of-content-creation/
You've heard about the streaming wars ( https://www.vccafe.com/2019/12/12/how-might-the-streaming-wars-escalate/), the cloud gaming/console wars ( https://www.vccafe.com/2020/09/21/the-rise-of-consumer-saas-and-the-bundle-wars/), the rise of the creator economy ( https://www.vccafe.com/2020/09/09/the-creator-economy-is-rising-but-challenges-abound/).
Behind all of these multi billion dollar markets vying for the consumer's time and money, content plays a key role. Quality content makes users join, buy, and stay subscribed.
But quality content is not easy to make or scale. There's the hits and blockbusters and then a very long tail of niche content looking for an audience. New technology developments in AI and graphic engines are changing that paradigm.
What if anyone with an idea could produce high quality content at scale? What if the burden of distribution was removed by clever recommendation engines as we've seen with TikTok?
For the first time, a few pieces of the puzzle are coming together to make high quality content easily created, distributed, consumed and monetised.
- Infrastructure - 4G is strong enough to serve streaming content and play video games in the cloud, but 5G will play a role in increasing content consumption, particularly video, AR/VR, volumetric video etc.
- Technology - Advancements in chips/GPUs enable making content that previously took hours, in second. Relatively new AI projects like GPT-3 or Gans are becoming more widely available, building upon years of research in NLP and computer vision.
- Connectivity - The number of smartphone users worldwide today surpasses three billion ( https://www.statista.com/statistics/330695/number-of-smartphone-users-worldwide/) and is expected to grow by hundreds of millions in 2021. We have access and consume more content than ever before in history.
With that backdrop, these five technologies are creating huge opportunities in content creation:
1. Gaming Engines
Speaking at GDC in 2017, Tim Sweeney, the CEO of Epic said ( https://www.gamesindustry.biz/articles/2017-04-13-sweeney-the-future-of-the-games-industry-make-everyone-a-creator): "The future of the games industry? Make everyone a creator”. That's perhaps what made Epic offer Unreal Engine for free, to dramatically reduce the barriers of entry for new content creators.
Unity, Unreal and Roblox are part of the majority of the games created in the $180 billion market. Gaming engines are now also powering new Hollywood shows and movies (The Mandalorian, Lion King, etc) as well as immersive VR content which is growing in popularity.
With an increased demand for streaming content, immersive content and gaming of all kinds, gaming engines are positioned to continue growing.
2. GPT-3
GPT, or “Generative Pre-Trained Transformer", is in simple terms, an AI text generator. The engine uses the input of content it fed, uses deep learning algorithms and logic to generate new text.
GPT-3 ( https://arxiv.org/abs/2005.14165 )is the latest version released by OpenAI ( https://openai.com/ ), an artificial intelligence lab backed by $1 billion dollars in funding from Microsoft.
Access to the GPT-3 API is not yet widely available, but developers are able to apply for a license from Microsoft.
In 2020, GPT-3 grabbed a lot of headlines for its high quality AI generated articles ( https://www.theguardian.com/commentisfree/2020/sep/08/robot-wrote-this-article-gpt-3), marketing materials, etc. For example, here's what an AI blog idea generator suggested for VC Cafe with the topic "Israeli startups" - not bad!
3. GANS
GANs ( https://en.wikipedia.org/wiki/Generative_adversarial_network), or "Generative Adversarial Networks" is a class of machine learning framework designed by Ian Goodfellow in 2014.
In simple terms, Gans are algorithmic architectures that use two neural networks, pitting one against the other (thus the “adversarial”) in order to generate new, synthetic instances of data that can pass for real data.
The machine can be fed a training set (say photos) to generate new photos that look real, but are completely synthetic. A good example of this is in images is This Person Does Not Exist ( https://thispersondoesnotexist.com/ ) by Nvidia.
Gans are the technology powering 'deepdakes'. While we hear a lot about the negative aspect of deep fakes including disinformation in politics or fake news, there's a huge potential in harnessing this kind of technology to create new content and improve user experience.
When it comes to content creation, a whole new industry has begun with Synthetic Media, starting from images and advancing into video.
Consumer applications for entertainment quickly became popular, for example Reface ( https://apps.apple.com/gb/app/reface-face-swap-videos/id1488782587), a Ukrainian startup that attracted $5.5 million from Andreessen Horowitz.
Disclaimer: Remagine Ventures ( http://www.remagineventures.com ) is an investor in HourOne ( https://www.hourne.ai ), the synthetic video creation platform based on real human characters.
4. DALL-E
Dall-E ( https://openai.com/blog/dall-e/ ), the latest project from OpenAI (released Jan 5 2020) creates images from text. It combines the capabilities of GPT-3 to generate text and Image GPT to generate images.
The result is realistic looking images like this 'avocado armchair' or 'pentagonal green clock'
Dell-E can also paint and illustrate (graphics). The possibilities are really exciting - but the technology is not perfect yet, and as is the case with Gans, responsible use is needed to prevent bad actors from using this to spread misinformation/fake news.
5. Text to Voice
By leveraging Deep learning and using a combination of voice to text and text to voice, AI enables Brad Pitt to speak in perfect Hebrew using his own voice or Samuel L Jackson to speak Spanish.
By leveraging Deep learning and using a combination of voice to text and text to voice, AI enables Brad Pitt to speak in perfect Hebrew using his own voice or Samuel L Jackson to speak Spanish.
This tech can be used for dubbing of existing content and is also used to create synthetic voices based on original voice sample. Israeli startup Deepdub ( https://venturebeat.com/2020/12/16/deepdub-uses-ai-to-localize-dubbing-for-foreign-language-films/) is doing some interesting work in this space.
The future of content creation is bright. But with new capabilities also come new threats - content moderation is increasingly hard for social platforms (from Facebook/Twitter/Tiktok to Parler) and misinformation is ripe in messaging - huge challenges that remain open.
If you're building a company in the space of content creation using cutting edge tech - I'd love to talk to you. This is an area we actively focus on at Remagine Ventures.
This thread can be read here: https://www.vccafe.com/2021/02/02/five-advancements-powering-the-future-of-content-creation/
