Most scientific advances arrive at a safe distance. When NASA released the first jaw-dropping photos from the James Webb Space Telescope last Summer, stargazers around the world marveled at distant supernovae without any fear of getting caught up in the cataclysm. Science was on the move, but it wasn’t at our doorstep.
Not so, with the latest generation of artificial intelligence (AI), surnamed “generative” for its uncanny ability to translate instructions into compelling text, images and more. When ChatGPT rocketed to more than 100 million users in two months – a dozen times faster than game changers like the iPhone or the internet – it was clear that some new future was already here.
The sudden arrival raises concerns for many. But this specific moment in AI discourse is also a powerful opportunity to shape AI toward the public good by investing in public resources to make it easier to build safe and inclusive technologies.
Top AI ML Updates: A Guide on How MLCopilot is Using LLMs & Empowering Developers to Straighten ML Processes
There are good reasons to fear the emergence of capable AI. As with previous industrial revolutions, AI has the potential to concentrate wealth and power, to upset the balance between nations or between law and disorder, to undermine human potential, to cement ongoing injustice or to force rapid change on a beleaguered working class, (including *gulp* each one of us.) But it is also worth reserving a moment for wonder and possibility.
Like NASA’s Telescope, the latest generation of AI represents a human achievement, expanding on what we knew to be possible. It is the product of many hands, working together in broad daylight through academic conferences and publicly available tools. Private businesses like Meta and Alphabet have (uncharacteristically) given away free AI software that required millions in electricity and hundreds of millions in research salaries to develop; a credit to America’s culture of competitive, public innovation.
New technologies typically launch into the hands of a few. The average tailor in the 1840s could scarcely afford a power loom. Human computers doing math by hand in the 1960s would wait decades more to own the digital machines that were displacing their occupation. But the floodgates to utilize ChatGPT, Bing, Bard and newer offerings are wide open. Anyone with an internet connection can experiment with the most advanced technology on Earth and consider ways future versions might impact their lives..
When Facebook passed 100M monthly users in 2009, there was no one yet qualified to study its impact, real or potential, on society. AI has arrived escorted by chaperones: academic communities and industrial teams doing hard work on health and safety, attribution, and alignment to human goals and values. The speed of debate about AI’s risks is evidence of the guardrails being laid in front of us.
Recommended: AI + Analytics Can Do Wonders- Key Takeaways from the ThoughtSpot Beyond 2023 Event
AI is experiencing a fragile Spring, where its rapid growth is happening in public view instead of being driven underground by fear and greed. How can we extend that Spring into a Summer where AI advances benefit all, and reflect the values of our society?
One clue lies in the fact that the latest AI advances are themselves the result of public resources. ImageNet, a public databank of images organized in 2009 by the prescient Dr. Fei Fei Li (now of Stanford), is widely recognized to have been an essential catalyst for the Deep Learning revolution that gave way to today’s groundbreaking LLMs. Today, an AI startup called HuggingFace (after a welcoming emoji) plays an outsized role in making data and code associated with AI research available to a broad community. But without a public mandate, these community investments are fragmented and vulnerable.
During the genomics revolution of the 1990s, the US National Laboratory system played a key role in administering GenBank, a public database of genetic information accessible to all researchers. Sharing data publicly became a de facto requirement for publication – along with consideration of Ethical, Legal and Social Implications (ELSI) of each piece of research. Countless medical advances, and a foundation for ethical genomic research, can be traced back to the coordinated research culture that this created.
Today, AI is at a similar crossroads. We have a window where investments in public data, software, and educational resources can have an outsized impact on keeping AI research transparent, inclusive, and ethical. If we want AI that is reflective of our values, it’s incumbent upon elected officials, government, academia and the private sector to come together to invest in public research that helps to reflect those values. Opening the AI Summer to as many people as possible is the best way of making sure that our children grow up looking at AI as a wondrous tool for their benefit, and not their competition.
Comments are closed.