It’s a bumper week for government pushback on the misuse of artificial intelligence.
Today the EU released its long-awaited set of AI regulations, an early draft of which leaked last week. The regulations are wide ranging, with restrictions on mass surveillance and the use of AI to manipulate people.
But a statement of intent from the US Federal Trade Commission, outlined in a short blog post by staff lawyer Elisa Jillson on April 19, may have more teeth in the immediate future. According to the post, the FTC plans to go after companies using and selling biased algorithms.
A number of companies will be running scared right now, says Ryan Calo, a professor at the University of Washington, who works on technology and law. “It’s not really just this one blog post,” he says. “This one blog post is a very stark example of what looks to be a sea change.”
The EU is known for its hard line against Big Tech, but the FTC has taken a softer approach, at least in recent years. The agency is meant to police unfair and dishonest trade practices. Its remit is narrow—it does not have jurisdiction over government agencies, banks, or nonprofits. But it can step in when companies misrepresent the capabilities of a product they are selling, which means firms that claim their facial recognition systems, predictive policing algorithms or healthcare tools are not biased may now be in the line of fire. “Where they do have power, they have enormous power,” says Calo.
The FTC has not always been willing to wield that power. Following criticism in the 1980s and ’90s that it was being too aggressive, it backed off and picked fewer fights, especially against technology companies. This looks to be changing.
In the blog post, the FTC warns vendors that claims about AI must be “truthful, non-deceptive, and backed up by evidence.”
“For example, let’s say an AI developer tells clients that its product will provide ‘100% unbiased hiring decisions,’ but the algorithm was built with data that lacked racial or gender diversity. The result may be deception, discrimination—and an FTC law enforcement action.”
The FTC action has bipartisan support in the Senate, where commissioners were asked yesterday what more they could be doing and what they needed to do it. “There’s wind behind the sails,” says Calo.
Meanwhile, though they draw a clear line in the sand, the EU’s AI regulations are guidelines only. As with the GDPR rules introduced in 2018, it will be up to individual EU member states to decide how to implement them. Some of the language is also vague and open to interpretation. Take one provision against “subliminal techniques beyond a person’s consciousness in order to materially distort a person’s behaviour” in a way that could cause psychological harm. Does that apply to social media news feeds and targeted advertising? “We can expect many lobbyists to attempt to explicitly exclude advertising or recommender systems,” says Michael Veale, a faculty member at University College London who studies law and technology.
It will take years of legal challenges in the courts to thrash out the details and definitions. “That will only be after an extremely long process of investigation, complaint, fine, appeal, counter-appeal, and referral to the European Court of Justice,” says Veale. “At which point the cycle will start again.” But the FTC, despite its narrow remit, has the autonomy to act now.
One big limitation common to both the FTC and European Commission is the inability to rein in governments’ use of harmful AI tech. The EU’s regulations include carve-outs for state use of surveillance, for example. And the FTC is only authorized to go after companies. It could intervene by stopping private vendors from selling biased software to law enforcement agencies. But implementing this will be hard, given the secrecy around such sales and the lack of rules about what government agencies have to declare when procuring technology.
Yet this week’s announcements reflect an enormous worldwide shift toward serious regulation of AI, a technology that has been developed and deployed with little oversight so far. Ethics watchdogs have been calling for restrictions on unfair and harmful AI practices for years.
The EU sees its regulations bringing AI under existing protections for human liberties. “Artificial intelligence must serve people, and therefore artificial intelligence must always comply with people’s rights,” said Ursula von der Leyen, president of the European Commission, in a speech ahead of the release.
Regulation will also help AI with its image problem. As von der Leyen also said: “We want to encourage our citizens to feel confident to use it.”
These robots know when to ask for help
A new training model, dubbed “KnowNo,” aims to address this problem by teaching robots to ask for our help when orders are unclear. At the same time, it ensures they seek clarification only when necessary, minimizing needless back-and-forth. The result is a smart assistant that tries to make sure it understands what you want without bothering you too much.
Andy Zeng, a research scientist at Google DeepMind who helped develop the new technique, says that while robots can be powerful in many specific scenarios, they are often bad at generalized tasks that require common sense.
For example, when asked to bring you a Coke, the robot needs to first understand that it needs to go into the kitchen, look for the refrigerator, and open the fridge door. Conventionally, these smaller substeps had to be manually programmed, because otherwise the robot would not know that people usually keep their drinks in the kitchen.
That’s something large language models (LLMs) could help to fix, because they have a lot of common-sense knowledge baked in, says Zeng.
Now when the robot is asked to bring a Coke, an LLM, which has a generalized understanding of the world, can generate a step-by-step guide for the robot to follow.
The problem with LLMs, though, is that there’s no way to guarantee that their instructions are possible for the robot to execute. Maybe the person doesn’t have a refrigerator in the kitchen, or the fridge door handle is broken. In these situations, robots need to ask humans for help.
KnowNo makes that possible by combining large language models with statistical tools that quantify confidence levels.
When given an ambiguous instruction like “Put the bowl in the microwave,” KnowNo first generates multiple possible next actions using the language model. Then it creates a confidence score predicting the likelihood that each potential choice is the best one.
The Download: inside the first CRISPR treatment, and smarter robots
The news: A new robot training model, dubbed “KnowNo,” aims to teach robots to ask for our help when orders are unclear. At the same time, it ensures they seek clarification only when necessary, minimizing needless back-and-forth. The result is a smart assistant that tries to make sure it understands what you want without bothering you too much.
Why it matters: While robots can be powerful in many specific scenarios, they are often bad at generalized tasks that require common sense. That’s something large language models could help to fix, because they have a lot of common-sense knowledge baked in. Read the full story.
Medical microrobots that travel inside the body are (still) on their way
The human body is a labyrinth of vessels and tubing, full of barriers that are difficult to break through. That poses a serious hurdle for doctors. Illness is often caused by problems that are hard to visualize and difficult to access. But imagine if we could deploy armies of tiny robots into the body to do the job for us. They could break up hard-to-reach clots, deliver drugs to even the most inaccessible tumors, and even help guide embryos toward implantation.
We’ve been hearing about the use of tiny robots in medicine for years, maybe even decades. And they’re still not here. But experts are adamant that medical microbots are finally coming, and that they could be a game changer for a number of serious diseases. Read the full story.
5 things we didn’t put on our 2024 list of 10 Breakthrough Technologies
We haven’t always been right (RIP, Baxter), but we’ve often been early to spot important areas of progress (we put natural-language processing on our very first list in 2001; today this technology underpins large language models and generative AI tools like ChatGPT).
Every year, our reporters and editors nominate technologies that they think deserve a spot, and we spend weeks debating which ones should make the cut. Here are some of the technologies we didn’t pick this time—and why we’ve left them off, for now.
New drugs for Alzheimer’s disease
Alzmeiher’s patients have long lacked treatment options. Several new drugs have now been proved to slow cognitive decline, albeit modestly, by clearing out harmful plaques in the brain. In July, the FDA approved Leqembi by Eisai and Biogen, and Eli Lilly’s donanemab could soon be next. But the drugs come with serious side effects, including brain swelling and bleeding, which can be fatal in some cases. Plus, they’re hard to administer—patients receive doses via an IV and must receive regular MRIs to check for brain swelling. These drawbacks gave us pause.
Sustainable aviation fuel
Alternative jet fuels made from cooking oil, leftover animal fats, or agricultural waste could reduce emissions from flying. They have been in development for years, and scientists are making steady progress, with several recent demonstration flights. But production and use will need to ramp up significantly for these fuels to make a meaningful climate impact. While they do look promising, there wasn’t a key moment or “breakthrough” that merited a spot for sustainable aviation fuels on this year’s list.
One way to counteract global warming could be to release particles into the stratosphere that reflect the sun’s energy and cool the planet. That idea is highly controversial within the scientific community, but a few researchers and companies have begun exploring whether it’s possible by launching a series of small-scale high-flying tests. One such launch prompted Mexico to ban solar geoengineering experiments earlier this year. It’s not really clear where geoengineering will go from here or whether these early efforts will stall out. Amid that uncertainty, we decided to hold off for now.