Artificial Intelligence is without any doubt one of the key themes of 2022 in technology. This year might well be remember as the year of painting ostentatiously. Leading by DALL-E, the AI model by OpenAI, diffusion models have taken over text-to-image generation and are rapidly expanding into other modalities, beginning with text-to-video generation.

ChatGPT, the newest OpenAI GPT-3.5 language model conversational bot, was also launched in November 2022, and has also garnered huge attention for its detailed responses and articulate answers. However, unlike with image generation, factual accuracy is still a challenge, the very same that forced Meta to retire its model Galactica.

Galactica demo is off line for now.
It's no longer possible to have some fun by casually misusing it.
Happy? https://t.co/K56r2LpvFD

— Yann LeCun (@ylecun) November 17, 2022

Recent developments offer us a number of interesting reflections. Design is becoming a writer’s medium, at least for the time being. The Future of Coding is Conversation. The Undergraduate Essay Is About to Die, and journalists… poor journalists.

Specifically, in 2022, investment in startups using AI has slowed down along with the broader market.

And adoption in the corporate world is a lot less exciting, with the optimization of service operations as the killer application.

Among the ~ 1,3 Billion results that Google offers when you search for “state of AI in 2022,” an interesting report is the one published by Nathan Benaich and Ian Hogarth (11 October 2022). This is their executive summary:

Research

• Diffusion models took the computer vision world by storm with impressive text-to-image generation capabilities.
• AI attacks more science problems, ranging from plastic recycling, nuclear fusion reactor control, and natural product discovery.
• Scaling laws refocus on data: perhaps model scale is not all that you need. Progress towards a single model to rule them all.
• Community-driven open sourcing of large models happens at breakneck speed, empowering collectives to compete with large labs.
• Inspired by neuroscience, AI research are starting to look like cognitive science in its approaches.

Industry

• Have upstart AI semiconductor startups made a dent vs. NVIDIA? Usage statistics in AI research shows NVIDIA ahead by 20-100x.
• Big tech companies expand their AI clouds and form large partnerships with A(G)I startups.
• Hiring freezes and the disbanding of AI labs precipitates the formation of many startups from giants including DeepMind and OpenAI.
• Major AI drug discovery companies have 18 clinical assets and the first CE mark is awarded for autonomous medical imaging diagnostics.
• The latest in AI for code research is quickly translated by big tech and startups into commercial developer tools.
  

Politics

• The chasm between academia and industry in large scale AI work is potentially beyond repair: almost 0% of work is done in academia.
• Academia is passing the baton to decentralized research collectives funded by non-traditional sources.
• The Great Reshoring of American semiconductor capabilities is kicked off in earnest, but geopolitical tensions are sky high.
• AI continues to be infused into a greater number of defense product categories and defense AI startups receive even more funding.
  

Safety

• AI Safety research is seeing increased awareness, talent, and funding, but is still far behind that of capabilities research.

And this is my favourite takeaway: Universities are a hotbed for AI spinouts, but a widening compute chasm is separating industry from academia in large model AI. Almost 0% of AI work is done in academia.

The baton is passing from academia to decentralized research collectives. Stability AI is attempting a new paradigm in commercializable open-source.

Stability AI, one of the contenders in text to image generation, announced the public release of Stable Diffusion in August, and entered the realm of unicorns in October.

The long term trend offers no doubt, at least for techno-optimistics.

Based on the steady advances in AI technology and the large recent increases in investment, we should expect AI technology to become even more powerful and impactful in the years and decades to come.

And the number of question that raises is amazing.

Will an AI be able to reliably construct bug-free code of more than 10,000 lines before 2030? Metaculus Community Prediction: 51%

Will AI be able to watch a movie and tell you accurately what is going on before 2030? Metaculus Community Prediction: 89%

(this is an interesting question in my head for years. Great expectation)

Imagine a world where you need not drive, you need not cook, you need not in fact look for information because a personal assistant is doing it all for you, recommending everything and curating everything. In fact, you might not need watch any film or read any book if you don’t want to because your assistant will be able to watch 100 films and read 1.000 books every day and can provide you with a summary or, eventually, give you a recommendation: You might read this paragraph and watch this scene, thought in fact of course you need not…

Will AI progress surprise us? Metaculus Community Prediction: 74%

You may find many other similar questions (and participate) HERE. The stakes are high!

And if you have not published your own report about “The State of AI 2022” yet, what are you waiting for? In case you are too busy, just asks ChatGPT… Oh wait.

____________________

Featured Image: Vibrant portrait painting of Salvador Dalí with a robotic half face, by DALL-E. (of which there are of course countless variations)