Industry executives and experts share their predictions for 2021. Read them in this 13th annual VMblog.com series exclusive.
AI progress in 2020 and predictions for 2021
By Miao He, an investor at Luminous Ventures
At Luminous Ventures, we believe artificial intelligence or AI has the potential to profoundly change society. However, there are still many cultural and organizational changes need to happen for us to see broad adoption of the technology. 2020 brought many technological breakthroughs but this is just the beginning. Here are our thoughts on what we think will happen in 2021.
Pushing the boundary of AI research
The year 2020 has witnessed major breakthroughs in AI such as GPT-3 for language and DeepMind’s AlphaFold for protein structure prediction. Each of these is a giant leap of historical importance and will most likely shape industries in a profound way. GPT-3 generates text using algorithms that are pre-trained and is better at creating content than any algorithm that has come before it. We have already been many interesting applications of what it can do. The effort of building more advanced models for language processing will continue next year, though it is also possible that the incremental gain compared to existing models will not be huge, as some researchers have pointed out that the progress in core AI innovations has been stagnant. Complex language models are costly to train and the most push on this front has been from the big techs or research institutes funded by them. OpenAI’s GPT-3 has 175 billion parameters and is said to have cost $10m to develop.
Organisational learning on top of machine learning
At the same time, more activities will happen within organisations to flesh out AI’s potential. While AI research continues to produce exciting results, still a lot more work is needed to turn them into practices that drive impact to the bottom line. According to a study released by MIT Sloan Management Review, so far only 10% of companies are achieving significant financial benefits with artificial intelligence. The research shows that these companies intentionally change processes, broadly and deeply to facilitate organisational learning with AI. As more organisations are waking up to the importance of ‘organisaional learning’ on top of ‘machine learning’, we believe more companies will shift from engaging in pilot studies and purchasing solutions to creating real organisational changes to integrate AI into their business processes. A research by McKinsey also showed that companies that are high AI performers tend to have the ability to build AI solutions in house. This means employing more AI talents and building a full-fledged team that consists of data scientists, data engineers, and application experts. We are seeing this happening in healthcare, where large pharma companies such as Novartis have been actively hiring data science talents. The competition for AI talents will continue in the next year, especially with the remote work trend, top talents will increasingly be sourced across wider geographies.
AI-driven drug discovery is proving its value
In healthcare we are more likely to see IPOs of companies that integrate AI into their drug discovery process. AbCellera, a Canadian biotechnology company with a full-stack AI-powered therapeutic antibody drug discovery platform IPO’ed this month and has reached $16 billion market cap at its peak. Its lead program – a novel antibody against COVID-19 co-developed with Lilly has received emergency use authorization from the FDA. This is a strong validation of its novel approach and represents great promise of accelerating drug discovery, which is traditionally long and costly. As large pharma players see more successes emerging from this space, it is possible that the bold ones will start to make acquisitions. Over the past years we have already seen a flurry of partnership deals between AI-first drug discovery companies and large pharma, such as insitro with Gilead, Recursion with Takeda, and Exscientia with Sanofi to name a few, with potential deal value of $1 billion or more for some of the deals. We see AI-driven drug discovery proving out its value and starting to make significant impact in the healthcare industry.
##
About the Author
Miao He is a scientist passionate about the potential of new technologies bringing changes to all aspects of our lives. Previously she worked as a Bioinformatics Scientist at Illumina, where she combined genetic and computational approaches to interpreting genomic sequencing data to help deliver better cancer management. At Illumina she worked with early members of Solexa and collaborated with key opinion leaders in academia and Genomics England on the 100,000 Genomes Project. Miao has consulted for Cancer Research UK and healthcare/biotech businesses and has worked in early-stage investing at OurCrowd, a venture capital firm in Israel. Miao has a Ph.D. in Molecular Biology from the University of Cambridge and an MBA from London Business School. Miao is the author of 14 peer-reviewed publications and has first-author papers in high-impact journals such as Nature Genetics. Her research has been featured in the Guardian, BBC News, LA Times etc.
