The Download: OpenAI’s plans for science, and chatbot age verification
Inside OpenAI's Big Play for Science
In the three years since ChatGPT's explosive debut, OpenAI's technology has upended a remarkable range of everyday activities at home, at work, and in schools. Now, OpenAI is making an explicit play for scientists. In October, the firm announced that it had launched a whole new team, called OpenAI for Science, dedicated to exploring how its large language models could help scientists and tweaking its tools to support them.
So why now? How does a push into science fit with OpenAI's wider mission? And what exactly is the firm hoping to achieve? I put these questions to Kevin Weil, a vice president at OpenAI who leads the new OpenAI for Science team, in an exclusive interview.
"We're excited to bring our technology to the scientific community," Weil says. "We believe that our models can help scientists with tasks such as data analysis, literature review, and even hypothesis generation. We're also working on developing new tools that can help scientists communicate their findings more effectively."
Weil notes that OpenAI's language models have already been used in various scientific applications, such as helping researchers analyze large datasets and generate new hypotheses. However, he emphasizes that the new OpenAI for Science team is focused on further developing and refining these tools to meet the specific needs of scientists.
"We're not just looking to provide a tool that can do a specific task," Weil explains. "We're looking to provide a platform that can help scientists collaborate, share knowledge, and accelerate the pace of scientific discovery."
One of the key challenges that OpenAI's language models face in the scientific community is the need for high-quality training data. Weil notes that scientists often have to spend a significant amount of time collecting and annotating data, which can be a barrier to entry for researchers who are not familiar with machine learning.
To address this challenge, OpenAI is working on developing new tools that can help scientists collect and annotate data more efficiently. Weil notes that the firm is also exploring the use of transfer learning, which involves pre-training a model on a large dataset and then fine-tuning it on a smaller dataset.
"We're excited about the potential of transfer learning to help scientists get started with machine learning more quickly," Weil says. "We believe that this can help level the playing field and make machine learning more accessible to researchers who may not have the resources or expertise to develop their own models from scratch."
Chatbots and Age Verification
How do tech companies check if their users are kids? This question has taken on new urgency recently thanks to growing concern about the dangers that can arise when children talk to AI chatbots. For years, Big Tech asked for birthdays (that one could make up) to avoid violating child privacy laws, but they weren't required to moderate content accordingly.
Now, two developments over the last week show how quickly things are changing in the US and how this issue is becoming a new battleground, even among parents and child-safety advocates.
In a recent interview, James O'Donnell, a tech journalist, noted that chatbots are starting to check users' ages to ensure they are not interacting with minors. This is a significant shift from the previous approach, where companies relied on users to provide their age voluntarily.
"We're seeing a growing recognition that chatbots need to be designed with safety in mind," O'Donnell says. "This includes ensuring that users are not interacting with minors, as well as providing clear guidelines and moderation policies to prevent the spread of harmful content."
O'Donnell notes that this shift is driven by a combination of factors, including increased awareness of the risks associated with chatbots and the growing demand for more robust safety measures.
TR10: Commercial Space Stations
Humans have long dreamed of living among the stars, and for two decades, hundreds of us have done so aboard the International Space Station (ISS). But a new era is about to begin in which private companies operate orbital outposts – with the promise of much greater access to space than before.
The ISS is aging and is expected to be brought down from orbit into the ocean in 2031. To replace it, NASA has awarded more than $500 million to several companies to develop private space stations, while others have built versions on their own.
We made them one of our 10 Breakthrough Technologies this year, and we're excited to explore the possibilities of commercial space stations.
The Must-Reads
Here are some of the most interesting and thought-provoking stories from around the web:
- Tech workers are pressuring their bosses to condemn ICE: The biggest companies and their leaders have remained largely silent so far. (Axios)
- The US Department of Transport plans to use AI to write new safety rules: Please don't do this. (ProPublica)
- The FBI is investigating Minnesota Signal chats tracking federal agents: But free speech advocates claim the information is legally obtained. (NBC News)
- TikTok users claim they're unable to send "Epstein" in direct messages: But the company says it doesn't know why. (NPR)
- Grok is not safe for children or teens: That's the finding of a new report digging into the chatbot's safety measures. (TechCrunch)
- The US is on the verge of losing its measles-free status: Following a year of extensive outbreaks. (Undark)
- Georgia has become the latest US state to consider banning data centers: Joining Maryland and Oklahoma's stance. (The Guardian)
- The future of Saudi Arabia's futuristic city is in peril: The Line was supposed to house 9 million people. Instead, it could become a data center hub. (FT $)
- Where do Earth's lighter elements go?: New research suggests they might be hiding deep inside its core. (Knowable Magazine)
- AI-generated influencers are getting increasingly surreal: Featuring virtual conjoined twins, and triple-breasted women. (404 Media)
Quote of the Day
"Humanity is about to be handed almost unimaginable power, and it is deeply unclear whether our social, political, and technological systems possess the maturity to wield it."
— Anthropic CEO Dario Amodei sounds the alarm about what he sees as the imminent dangers of AI superintelligence in a new 38-page essay, Axios reports.
One More Thing
Why one developer won't quit fighting to connect the US's grids: Michael Skelly hasn't learned to take no for an answer. For much of the last 15 years, the energy entrepreneur has worked to develop long-haul transmission lines to carry wind power across the Great Plains, Midwest, and Southwest. But so far, he has little to show for the effort.
Skelly has long argued that building such lines and linking together the nation's grids would accelerate the shift from coal- and natural-gas-fueled power plants to the renewables needed to cut the pollution driving climate change. But his previous business shut down in 2019, after halting two of its projects and selling off interests in three more.
Skelly contends he was early, not wrong. And he has a point: markets and policymakers are increasingly coming around to his perspective.
