AI talk: Patent lawyer and a 2024 preview

Jan. 12, 2024 / By V. “Juggy” Jagannathan, PhD

This week it is my privilege to share a podcast I did with Robert Plotkin. Robert is a patent lawyer who has been exploring the role of artificial intelligence (AI) in inventions for more than a decade. I’ll round off this 2024 inaugural blog with a sneak peek at what we can expect in terms of large language models (LLMs). 

Patently an AI patent lawyer 

Robert Plotkin was an M*Modal software patent attorney before 3M acquired M*Modal in 2019. In 2009 he published the book, “The Genie in the Machine,” an exploration of the role of AI in creating inventions. He co-founded Blueshift IP and is actively supporting software patents using AI. He is poised to release his second book in this arena in a few months. Robert and I explore a range of patent related topics in this podcast, and I came away impressed with the depth and breadth of his understanding of the current generative AI tech and its role in creating patentable inventions. Listen to our conversation here. 

Hands hovering around a crystal ball. 

Generative AI in 2024 

The year 2023 will go down as the year generative AI went mainstream! We saw eye-popping progress in the use of LLMs. So, what’s on the horizon for 2024? Of course, if you Google it, there are predictably lot of predictions on this front, from MIT Tech Report to Forbes and others. Let’s sample some of these.

MIT Tech Forum explored this very issue in the article, “What’s next for AI in 2024” Included are four predictions:  

  1.  Chatbots will be customized with little to no coding needed 
  2. Text-to-video generation will take off  
  3. AI generated disinformation will cause chaos with elections
  4. Robots will learn to multitask  

My take on these predictions? Chatbots are already being customized, sometimes with unintended consequences – as it is not easy to control the output of generative AI (yet). An AI film festival is in the works to judge the quality of short videos generated by AI – an indication of where video generation is headed. The worrying prediction about disinformation, I hope that does not come to pass. We don’t need any more chaos in this or any election year. Self-driving cars are getting a new boost of life. DeepMind keeps releasing new robo foundation models.

Forbes came up with 10 predictions. I will just highlight a few here: 

  1. LLMs will give way to large multi-modal models (LMMs) or more specialized foundational models. This is already happening with both Google and Open AI now supporting multiple modalities. Closed models from Open AI, Google, Anthropic, etc., will continue to outperform open-source models. This is likely to be true, though a number of open-source solutions will probably be good enough for a variety of specialized applications.
  2. Alternatives to transformer architecture will gain traction. A few surfaced last year – whether it will see the billions of dollars to take it to the level of current transformer-based LLMs, remains to be seen.
  3. There is likely to be a resolution of the various copyright infringement lawsuits against Open AI and others in the use of copyrighted material for training the large models. Any ruling on this front can be consequential – which hinges around the term “fair use.” 

I saw a LinkedIn post by Meta and endorsed by their chief AI scientist Yann Lecun, with their own take on 2024. Not surprisingly, it had a Meta-slant, including a plug for Meta’s RayBan smart glasses. The company confirmed the sentiment that LLMs will give way to LMMs. Another of the predictions: open-source models will beat GPT4. Llama 2 derivatives and Mistral AI have closed the gap but not overtaken the leading models yet. Purpose built small language models (SLMs) will continue to be built and deployed.

One last set of predictions that I am going to include here is from Stanford HAI 

  1. A shift in white collar jobs. AI is going to enhance productivity of a good range of white collar workers.
  2. LMMs producing believable deep fakes.
  3. A proliferation of AI use cases contributing to the GPU shortage. Peter Norvig, popular for his textbook on AI used across universities, predicts that the AI assistants (agents) will do more practical and useful stuff this year. 

Consumer Electronics Show (CES) 2024 

Predicting the tech landscape for 2024 would not be complete without paying homage to the happenings at the annual CES event. At the beginning of every year, tech nerds from around the planet congregate to binge on the future of tech. This year is no different. This year’s theme? What else? Generative AI! More than 130,000 people are congregating this week in Las Vegas, including 4,000 companies exhibiting tech offerings, with more than 1,200 of those being startups. Interestingly, this is the 100-year anniversary for the Consumer Technology Association which hosts CES.

All kinds of tech are on display: fin tech, food tech, health tech, pet tech, auto tech, smart cities, smart displays, etc. Al is being infused liberally across the board. The CEO of L’Oreal Groupe presented the tech behind beauty products and unsurprisingly, a chatbot was providing makeup advice. There is an AI enabled toothbrush, a ring that could pass for jewelry supporting all kinds of remote monitoring and women’s health metrics. Kids’ books and cartoons come alive with a chatbot interface. Robots roam the hospital corridor assisting others, apparently now these are called “cobots.” There is even a Fitbit for dogs! I noticed that they had a moderated session discussing what we can expect in 2024 with Fei-Fei Li and Andrew Ng. Their predictions are very much in line with what I discussed above.

Whatever the predictions from all these different entities, one thing is quite certain. We will witness another tumultuous year when it comes to AI.

 I am always looking for feedback and if you would like me to cover a story, please let me know! Leave me a comment below or ask a question on my blogger profile page. 

“Juggy” Jagannathan, PhD, is an AI evangelist with four decades of experience in AI and computer science research.