As summer winds down and the familiar hum of school buses returns to our neighborhoods, millions of American students are gearing up for another year of learning. But as we stand on the cusp of an artificial intelligence (AI) revolution, this annual ritual is about to face a seismic shift—especially for students with intellectual and developmental disabilities (IDD).

[time-brightcove not-tgx=”true”]

The decisions that school leaders make in the next academic year are likely to determine whether this technological wave creates more inclusive learning environments, or exacerbates existing disparities. A recent study from the Special Olympics Global Center for Inclusion in Education reveals a complex landscape of attitudes towards AI in education and a fear of leaving students with IDD behind.

The study found the majority of educators (64%) and parents (77%) of students with IDD view AI as a potentially powerful mechanism to promote more inclusive learning. AI will never replace the centrality of genuine human connection in teaching, the essential element for our community to flourish in the classroom or on the playing field. But contrary to the alarms that many are raising about AI in schools, our research demonstrated significant optimism about the technology. Those who work most closely with young people with ID see great potential in AI’s ability to simplify information—including lectures and curricula—making it more accessible to students with disabilities. Imagine the adaptive learning systems that can provide each student with an educational approach tailored to their unique needs.

But despite teachers’ confidence in AI’s potential for students with IDD, our research also shows that their fears about potential negative impacts on the general student population overshadow their enthusiasm for its role as a learning aid. Specifically, the majority of teachers (78%) express concern that the use of AI in schools might lead to a decrease in human interaction in schools, with 65% also worried about AI use potentially reducing students’ ability to practice empathy.

How do we overcome those fears? We found that teachers who have used AI are much more likely to think it can make education more inclusive, inspiring more creative thinking about how it can support their students with IDD. These educators are less prone to generalizing concerns that AI will negatively impact the classroom experience for the broader student population. Such findings demonstrate the importance of comprehensive teacher training on AI platforms. By familiarizing educators with AI tools, we can bridge the gap between potential and application, fostering more inclusive learning environments for all students.

But ultimately, educators’ experience with the tools alone is insufficient. Our study reveals concerns among teachers (72%) and parents (63%) that AI models themselves have not been trained on data provided by persons with IDD, and therefore do not accurately reflect their capabilities and contributions.

Thus, people with IDD must have a seat at the table when discussing the responsible use of AI in education. For example, Microsoft acknowledges that “humanity traverses a broad neural spectrum” and has expanded design approaches that once narrowly focused on physical disabilities to include differences in cognitive issues and learning styles. They have also acknowledged the importance of diversifying the teams that build and test AI, as well as making a conscious effort to identify bias in the data sets used to train AI systems. Greater attention to and adoption of inclusive design principles for educational technology will translate into more inclusive learning environments.  

A failure to listen to people with IDD will result in 3% of the population being locked out of the most revolutionary technology since the advent of the personal computer. That must not happen.

Making AI tools inclusive requires a collaborative effort among teachers, parents, and most importantly, tech companies. Special Olympics is calling on the companies that develop AI systems to convene experts in technology and inclusive education to initiate meaningful dialogues with the IDD community to ensure their needs and perspectives are considered in product development.

Many Special Olympics athletes are already using AI tools independently to help ensure they are understanding the nuances of conversations, meetings and lectures, as well as to help organize ever more complex schedules and training regimes. Education leaders can utilize AI to help identify when their school climate is sliding into toxicity, allowing for more targeted early interventions to address isolation and bullying. Young people with ID have been on the frontlines of this fight, working to create a climate of justice and joy where social inclusion is the norm. They have reflected deeply on AI and what it means for their future; we only need to ask them to share their thoughts.   

As pencils are sharpened and backpacks are filled, let’s also sharpen our resolve to make this school year a turning point for inclusive education. If used responsibly, AI can help to tear down the physical walls at segregated schools, as well as the invisible barriers that separate children of different abilities within the same classroom. We believe that AI can get us there.

Our research illuminates both the promise and the challenges of AI in supporting students with IDD. But it also shows us the way forward: through teacher training, community involvement, and a commitment to inclusive design. As we step into this new school year, we have before us an unprecedented opportunity to reshape education, to close gaps, and to unlock the full potential of every student. 

The school of tomorrow is being built today—let’s make sure it has room for everyone.

Few places in the world have more to gain from a flourishing AI industry than California. Few also have more to lose if the public’s trust in the industry were suddenly shattered.

In May, the California Senate passed SB 1047, a piece of AI safety legislation, in a vote of 32 to one, helping ensure the safe development of large-scale AI systems through clear, predictable, common-sense safety standards. The bill is now slated for a state assembly vote this week and, if signed into law by Governor Gavin Newsom, would represent a significant step in protecting California citizens and the state’s burgeoning AI industry from malicious use.

[time-brightcove not-tgx=”true”]

Late Monday, Elon Musk shocked many by announcing his support for the bill in a post on X. “This is a tough call and will make some people upset, but, all things considered, I think California should probably pass the SB 1047 AI safety bill,” he wrote. “For over 20 years, I have been an advocate for AI regulation, just as we regulate any product/technology that is a potential risk to the public.”

The post came days after I spoke with Musk about SB 1047. Unlike other corporate leaders who often waver, consulting their PR teams and lawyers before taking a stance on safety legislation, Musk was different. After I outlined the importance of the bill, he requested to review its text to ensure its fairness and lack of potential for abuse. The next day he came out in support. This quick decision-making process is a testament to Musk’s long-standing advocacy for responsible AI regulation.

Last winter, Senator Scott Weiner, the bill’s creator, reached out to the Center for AI Safety (CAIS) Action Fund for technical suggestions and cosponsorship. As CAIS’s founder, my commitment to transformative technologies impacting public safety is our mission’s cornerstone. To preserve innovation, we must anticipate potential pitfalls, because an ounce of prevention is worth a pound of cure. Recognizing SB 1047’s groundbreaking nature, we were thrilled to help and have advocated for its adoption ever since.

Read More: Exclusive: California Bill Proposes Regulating AI at State Level

Targeted at the most advanced AI models, it will require large companies to test for hazards, implement safeguards, ensure shutdown capabilities, protect whistleblowers, and manage risks. These measures aim to prevent cyberattacks on critical infrastructure, bioengineering of viruses, or other malicious activities with the potential to cause widespread destruction and mass casualties

Anthropic recently warned that AI risks could emerge in “as little as 1-3 years,” disputing critics who view safety concerns as imaginary. Of course, if these risks are indeed fictitious, developers shouldn’t fear liability. Moreover, developers have pledged to tackle these issues, aligning with President Joe Biden’s recent executive order, reaffirmed at the 2024 AI Seoul Summit.

Enforcement is lean by design, allowing California’s Attorney General to act only in extreme cases. There are no licensing requirements for new models, nor does it punish honest mistakes or criminalize open sourcing—the practice of making software source code freely available. It wasn’t drafted by Big Tech or those focused on distant future scenarios. The bill aims to prevent frontier labs from neglecting caution and critical safeguards in their rush to release the most capable models.

---

Like most AI safety researchers, I am in large part driven by a belief in its immense potential to benefit society, and deeply concerned about preserving that potential. As a global leader in AI, California is too. This shared concern is why state politicians and AI safety researchers are enthusiastic about SB 1047, as history tells us that a major disaster, like the nuclear one at Three Mile Island on March 28, 1979, could set a burgeoning industry back decades.

Regulatory bodies responded to the partial nuclear meltdown by overhauling nuclear safety standards and protocols. These changes increased the operational costs and complexity of running nuclear plants, as operators invested in new safety systems and complied with rigorous oversight. The regulatory challenges made nuclear energy less appealing, halting its expansion over the next 30 years.

Three Mile Island led to a greater dependence on coal, oil, and natural gas. It is often argued that this was a significant lost opportunity to advance toward a more sustainable and efficient global energy infrastructure. While it remains uncertain whether stricter regulations could have averted the incident, it is clear that a single event can profoundly impact public perception, stifling the long-term potential of an entire industry.

Some people will view any government action on industry with suspicion, considering it inherently detrimental to business, innovation, and a state or country’s competitive edge. Three Mile Island demonstrates this perspective is short-sighted, as measures to reduce the chances of a disaster are often in the long-term interest of emerging industries. It is also not the only cautionary tale for the AI industry.

When social media platforms first emerged, they were largely met with enthusiasm and optimism. A 2010 Pew Research Center survey found that 67% of American adults who used social media believed it had a mostly positive impact. Futurist Brian Solis captured this ethos when he proclaimed, “Social media is the new way to communicate, the new way to build relationships, the new way to build businesses, and the new way to build a better world.”

He was three-fourths correct.

Driven by concerns over privacy breaches, misinformation, and mental health impacts, public perception of social media has flipped, with 64% of Americans viewing it negatively. Scandals like Cambridge Analytica eroded trust, while fake news and polarizing content highlighted social media’s role in societal division. A Royal Society for Public Health study showed 70% of young people experienced cyberbullying, with 91% of 16-24-year-olds stating social media harms their mental wellbeing. Users and policymakers around the globe are increasingly vocal about needing stricter regulations and greater accountability from social media companies.

This did not happen because social media companies are uniquely evil. Like other emerging industries, the early days were a “wild west” where companies rushed to dominate a burgeoning market and government regulation was lacking. Platforms with addictive, often harmful content thrived, and we are now all paying the price. The companies—increasingly mistrusted by consumers and in the crosshairs of regulators, legislators, and courts—included.

The optimism surrounding social media wasn’t misplaced. The technology did have the potential to break down geographical barriers and foster a sense of global community, democratize information, and facilitate positive social movements. As the author Erik Qualman warned, “We don’t have a choice on whether we do social media, the question is how well we do it.”

The lost potential of social media and nuclear energy was tragic, but it’s nothing compared to squandering AI’s potential. Smart legislation like SB 1047 is our best tool for preventing this while protecting innovation and competition.

The history of technological regulation showcases our capacity for foresight and adaptability. When railroads transformed 19th-century transportation, governments standardized track gauges, signaling, and safety protocols. The advent of electricity led to codes and standards preventing fires and electrocutions. The automobile revolution necessitated traffic laws and safety measures like seat belts and airbags. In aviation, bodies like the FAA established rigorous safety standards, making flying the safest form of transportation.

History can only provide us with lessons. Whether to heed them is up to us.

Read 7 remaining paragraphs | Comments

Is the AI bubble about to burst? Every day that the stock prices of semiconductor champion Nvidia and the so-called “Fab Five” tech giants (Microsoft, Apple, Alphabet, Amazon, and Meta) fail to regain their mid-year peaks, more people ask that question. 

It would not be the first time in financial history that the hype around a new technology led investors to drive up the value of the companies selling it to unsustainable heights—and then get cold feet. Political uncertainty around the U.S. election is itself raising the probability of a sell-off, as Donald Trump expresses his lingering resentments against the Big Tech companies and his ambivalence towards Taiwan, where the semiconductors essential for artificial intelligence mostly get made. 

[time-brightcove not-tgx=”true”]

The deeper question is whether AI can deliver the staggering long-term value that the internet has. If you invested in Amazon in late 1999, you would have been down over 90% by early 2001. But you would be up over 4,000% today. 

A chorus of skeptics now loudly claims that AI progress is about to hit a brick wall. Models such as GPT-4 and Gemini have already hoovered up most of the internet’s data for training, the story goes, and will lack the data needed to get much smarter.

Read More: 4 Charts That Show Why AI Progress Is Unlikely to Slow Down

However, history gives us a strong reason to doubt the doubters. Indeed, we think they are likely to end up in the same unhappy place as those who in 2001 cast aspersions on the future of Jeff Bezos’s scrappy online bookstore. 

The generative AI revolution has breathed fresh life into the TED-ready aphorism “data is the new oil.” But when LinkedIn influencers trot out that 2006 quote by British entrepreneur Clive Humby, most of them are missing the point. Data is like oil, but not just in the facile sense that each is the essential resource that defines a technological era. As futurist Ray Kurzweil observes, the key is that both data and oil vary greatly in the difficulty—and therefore cost—of extracting and refining them.

Some petroleum is light crude oil just below the ground, which gushes forth if you dig a deep enough hole in the dirt. Other petroleum is trapped far beneath the earth or locked in sedimentary shale rocks, and requires deep drilling and elaborate fracking or high-heat pyrolysis to be usable. When oil prices were low prior to the 1973 embargo, only the cheaper sources were economically viable to exploit. But during periods of soaring prices over the decades since, producers have been incentivized to use increasingly expensive means of unlocking further reserves.

The same dynamic applies to data—which is after all the plural of the Latin datum. Some data exist in neat and tidy datasets—labeled, annotated, fact-checked, and free for download in a common file format. But most data are buried more deeply. Data may be on badly scanned handwritten pages; may consist of terabytes of raw video or audio, without any labels on relevant features; may be riddled with inaccuracies and measurement errors or skewed by human biases. And most data are not on the public internet at all.

Read More: The Billion-Dollar Price Tag of Building AI

An estimated 96% to 99.8% of all online data are inaccessible to search engines—for example, paywalled media, password-protected corporate databases, legal documents, and medical records, plus an exponentially growing volume of private cloud storage. In addition, the vast majority of printed material has still never been digitized—around 90% for high-value collections such as the Smithsonian and U.K. National Archives, and likely a much higher proportion across all archives worldwide.

Yet arguably the largest untapped category is information that’s currently not captured in the first place, from the hand motions of surgeons in the operating room to the subtle expressions of actors on a Broadway stage.

For the first decade after large amounts of data became the key to training state-of-the-art AI, commercial applications were very limited. It therefore made sense for tech companies to harvest only the cheapest data sources. But the launch of Open AI’s ChatGPT in 2022 changed everything. Now, the world’s tech titans are locked in a frantic race to turn theoretical AI advances into consumer products worth billions. Many millions of users now pay around $20 per month for access to the premium AI models produced by Google, OpenAI, and Anthropic. But this is peanuts compared to the economic value that will be unlocked by future models capable of reliably performing professional tasks such as legal drafting, computer programming, medical diagnosis, financial analysis, and scientific research. 

The skeptics are right that the industry is about to run out of cheap data. As smarter models enable wider adoption of AI for lucrative use cases, however, powerful incentives will drive the drilling for ever more expensive data sources—the proven reserves of which are orders of magnitude larger than what has been used so far. This is already catalyzing a new training data sector, as companies including Scale AI, Sama, and Labelbox specialize in the digital refining needed to make the less accessible data usable.

Read More: OpenAI Used Kenyan Workers on Less Than $2 Per Hour to Make ChatGPT Less Toxic

This is also an opportunity for data owners. Many companies and nonprofits have mountains of proprietary data that are gathering dust today, but which could be used to propel the next generation of AI breakthroughs. OpenAI has already spent hundreds of millions of dollars licensing training data, inking blockbuster deals with Shutterstock and the Associated Press for access to their archives. Just as there was speculation in mineral rights during previous oil booms, we may soon see a rise in data brokers finding and licensing data in the hope of cashing in when AI companies catch up.

Much like the geopolitical scramble for oil, competition for top-quality data is also likely to affect superpower politics. Countries’ domestic privacy laws affect the availability of fresh training data for their tech ecosystems. The European Union’s 2016 General Data Protection Regulation leaves Europe’s nascent AI sector with an uphill climb to international competitiveness, while China’s expansive surveillance state allows Chinese firms to access larger and richer datasets than can be mined in America. Given the military and economic imperatives to stay ahead of Chinese AI labs, Western firms may thus be forced to look overseas for sources of data unavailable at home.

Yet just as alternative energy is fast eroding the dominance of fossil fuels, new AI development techniques may reduce the industry’s reliance on massive amounts of data. Premier labs are now working to perfect techniques known as “synthetic data” generation and “self-play,” which allow AI to create its own training data. And while AI models currently learn several orders of magnitude less efficiently than humans, as models develop more advanced reasoning, they will likely be able to hone their capabilities with far less data.

There are legitimate questions about how long AI’s recent blistering progress can be sustained. Despite enormous long-term potential, the short-term market bubble will likely burst before AI is smart enough to live up to the white-hot hype. But just as generations of “peak oil” predictions have been dashed by new extraction methods, we should not bet on an AI bust due to data running out.