As artificial intelligence continues to rapidly evolve, Stanford’s AI Playground provides a convenient environment for faculty, staff, and students to experiment with AI technology.
Earlier this year, University IT began piloting the AI Playground, a Stanford-hosted environment that allows users to test out and compare a range of AI tools, including ChatGPT, Google Gemini, and Wolfram. As of Oct. 1, the AI Playground is accessible to all Stanford faculty, staff, students, postdocs, and visiting scholars.
The project to create the AI Playground stemmed from community feedback and the need for a centralized, accessible platform where community members could safely try AI tools, explained Chief Information Officer Steve Gallagher.
“Our goal was to put the tools into people’s hands in a more accessible way,” Gallagher said. “One of UIT’s strategic values is to innovate with purpose, and in the case of the Playground, we want to help enable the innovation happening across campus.”
The AI Playground is designed to be both an entry point into AI for users who are unsure where to start and a place of experimentation for those more advanced in AI usage. “The vision is to increase the collective literacy around artificial intelligence and these tools across campus,” Gallagher said.
The AI Playground is part of Stanford’s broader effort to address the growing role of AI in research, education, and administration. Separately, the AI at Stanford Advisory Committee, appointed by the provost to assess potential policy gaps for AI use at the university, recently released its initial findings.
Time to play
The AI Playground includes various large language models (LLMs), a form of generative AI specializing in text-based content. Users can access additional LLM plugins for image generation, web scraping, and AI-assisted Google services.
This versatility enables users to compare how different AI models handle the same task or use different models simultaneously to complete tasks such as document analysis, creating graphs and websites, making recommendations, or skill assessments.
UIT provides resources on how to use AI responsibly, how to use the AI Playground, and how to use generative AI tools, which are built using algorithms that can generate text, images, video, audio, and 3D models in response to user prompts.
Joshua Barnett, IT director and enterprise infrastructure architect, emphasized the importance of expanding AI access to more students.
“We needed to level the playing field and make it free to use,” said Barnett, who helped assemble the Playground. “Some faculty have started mandating its use in their classes, and so we moved up access to students to allow for the start of fall quarter.”
Most users at this time are curious how they can use AI in their existing workflow, studies, and research but are unsure where to begin, Barnett said. “A lot of people don’t know what they don’t know,” he said, adding that UIT plans to introduce more demos showcasing the platforms’ capabilities.
For example, a history professor and students used AI in one class to quickly analyze periodicals and handwritten journals. Staff have also leveraged AI to analyze spreadsheets or to draft emails. “It’s a good tool to help augment what we’re doing day to day and make us do our jobs better and more efficiently,” Barnett said.
However, Barnett said it’s important for users to remain vigilant and remember that LLMs can “hallucinate” or introduce errors; he reminded users to always verify anything AI produces.
“If you don’t like the output of an image or report, or the information that’s given to you is questionable, have a conversation with the language model and ask why it’s doing that and give it feedback,” Barnett said. “Ask it to cite sources. They do make stuff up sometimes, but they are a great start if you don’t know where to begin.”
Stanford’s AI Playground enables data privacy by taking advantage of Stanford’s infrastructure and vendor partnerships. Files uploaded to the Playground are not shared externally or used to train models.
The name “AI Playground” underscores its purpose as a space to safely experiment with and get comfortable using AI tools, said Ganesh Karkala, associate vice president for enterprise technology and chief technology officer. “We wanted people to play with their data set and ask a set of questions without worrying about how vendors will use this data,” Karkala said.
The university has agreements that ensure data remains in the university environment, and Stanford is also considering a data retention policy to automatically purge data after a set period.
Nevertheless, users should only share low- and moderate-risk data in the AI Playground, meaning they should not upload files with proprietary or private data. “A lot of day-to-day tasks can be covered under those low- or moderate-risk labels,” Barnett said. “Hopefully, in the near future we’ll be able to expand it all the way up to high-risk data.”
In the meantime, Stanford Medicine users should still use Stanford Medicine’s Secure GPT tool.
A balance
Russ Altman is the chair of the AI at Stanford Advisory Committee, which has established principles for AI use at the university, including the necessity for human oversight.
While there are legitimate concerns regarding AI, it’s important for Stanford to encourage people to learn about it and how it may improve people’s lives, said Altman, the Kenneth Fong Professor in the School of Engineering and professor of bioengineering, of genetics, of medicine–biomedical informatics research, and of biomedical data science, and a senior fellow at the Stanford Institute for Human-Centered Artificial Intelligence.
“There’s a huge education required of faculty, students, and staff about what it can do,” Altman continued. “I’ve met faculty who have not tried ChatGPT or any LLM yet, and as a faculty member you need to be aware of how AI can be used. The Stanford way is to take new technologies and balance encouragement of experimentation with acknowledgement of risks.”
Users can join the #ai-playground Slack channel to pose questions and share their experiences using the AI Playground. In January, UIT is also launching “AI Playground 101,” an interactive UIT Tech Training class.
“Stanford University, officially Leland Stanford Junior University, is a private research university in Stanford, California. The campus occupies 8,180 acres, among the largest in the United States, and enrols over 17,000 students.”
Please visit the firm link to site
