Cutting planet-warming pollution to near-zero will take more than inventing new clean technologies—it will require changing how the world invests in them. That’s especially true for industries like aviation, where developing and adopting greener solutions is risky and expensive, according to a University of California San Diego commentary piece in Science.

The paper calls for smarter ways of managing investment risk that could help speed up the shift toward cleaner air travel and other hard-to-decarbonize sectors.

“The aviation sector—a fast-growing source of greenhouse gases—illustrates the broader challenge of industrial decarbonization: too little investment in technologies that could yield the biggest climate benefits,” said the paper’s co-author David G. Victor, professor of innovation and public policy at the UC San Diego School of Global Policy and Strategy and co-director of the Deep Decarbonization Initiative.

The piece outlines a new approach that could help guide a coalition of research and development (R&D) programs alongside investors and airlines seeking to deploy new technologies to curb carbon emissions from the aviation industry.

“Despite all the chaos in global geopolitics and climate policies these days, there are large and growing pools of capital willing to take risks on clean technology,” Victor said. “What’s been missing is a framework to guide that capital to the riskiest but most transformative investments.”

He added that investors and research managers tend to focus on familiar, lower-risk projects like next-generation jet engines or recycled-fuel pathways.

“But getting aviation and other hard-to-abate sectors to near-zero emissions means taking on bigger risks with technologies and new lines of business that will be highly disruptive to the existing industry. Investors and airlines need to find smarter ways to encourage and manage these disruptive investments,” Victor said.

In the article, Victor and co-authors call for a more realistic framework to guide both research funding and private investment.

They propose a tool called an Aviation Sustainability Index (ASI)—a quantitative method to assess how different technologies or investments could help decouple emissions from growth in air travel.

The approach is designed to help investors distinguish between projects that only modestly improve efficiency and those that could significantly transform the sector’s climate impact.

The authors note that while roughly $1 trillion is expected to flow into aviation over the next decade, most of that money will simply make aircraft slightly more efficient. Few investors, they argue, have clear incentives to back the kind of breakthrough technologies—such as hydrogen propulsion, advanced aircraft designs, or large-scale sustainable fuel systems—that could substantially reduce emissions.

“Cleaner flight is possible, but it requires changing how we think about both risk and return,” Victor said. “We need new institutions, incentives, and partnerships that reward innovation, not just incrementalism.”

The commentary, written by a multinational team of scholars, also highlights a broader lesson for climate policy: global decarbonization goals such as “net zero by 2050” sound bold and ambitious. But when it becomes clear that they can’t be met, these goals make it harder to focus on the practical steps needed today to drive change in real-world markets.

Ultimately, the paper argues for action that begins now. By developing better tools to evaluate climate-friendly investments and by rewarding companies willing to take calculated risks on breakthrough technologies, governments, investors and industry leaders can accelerate real progress toward decarbonization.

The paper was co-authored by Thomas Conlon of University College Dublin, Philipp Goedeking of Johannes Gutenberg University of Mainz (Germany) and Andreas W. Schäfer of University College London.

Imagine you’re walking your dog. It interacts with the world around you—sniffing some things, relieving itself on others. You walk down the Embarcadero in San Francisco on a bright sunny day, and you see the Ferry Building in the distance as you look out into the bay. Your dog turns to you, looks you in the eye, and says, “Did you know this waterfront was blocked by piers and a freeway for 100 years?”

OK now imagine your dog looks like an alien and only you can see it. That’s the vision for a new capability created for the Niantic Labs AR experience Peridot.

Niantic, also the developer of the worldwide AR behemoth Pokémon Go, hopes to build out its vision of extending the metaverse into the real world by giving people the means to augment the space around them with digital artifacts. Peridot is a mobile game that lets users customize and interact with their own little Dots—dog-sized digital companions that appear on your phone’s screen and can look like they’re interacting with the world objects in the view of your camera lens. They’re very cute, and yes, they look a lot like Pokémon. Now, they can talk.

Peridot started as a mobile game in 2022, then got infused with generative AI features. The game has since moved into the hands of Niantic Spatial, a startup created in April that aims to turn geospatial data into an accessible playground for its AR ambitions. Now called Peridot Beyond, it has been enabled in Snap’s Spectacles.

Hume AI, a startup running a large language model that aims to make chatbots seem more empathetic, is now partnering with Niantic Spatial to bring a voice to the Dots on Snap’s Spectacles. The move was initially announced in September, but now it’s ready for the public and will be demonstrated at Snap’s Lens Fest developer event this week.

Say a person takes their French Bulldog, Bowser, to the dog park. Identifying Bowser as he plays among the other canines is easy for the dog owner to do while onsite.

But if someone wants to use a generative AI model like GPT-5 to monitor their pet while they are at work, the model could fail at this basic task. Vision-language models like GPT-5 often excel at recognizing general objects, like a dog, but they perform poorly at locating personalized objects, like Bowser the French Bulldog.

To address this shortcoming, researchers from MIT and the MIT-IBM Watson AI Lab have introduced a new training method that teaches vision-language models to localize personalized objects in a scene.

Their method uses carefully prepared video-tracking data in which the same object is tracked across multiple frames. They designed the dataset so the model must focus on contextual clues to identify the personalized object, rather than relying on knowledge it previously memorized.

When given a few example images showing a personalized object, like someone’s pet, the retrained model is better able to identify the location of that same pet in a new image.

Models retrained with their method outperformed state-of-the-art systems at this task. Importantly, their technique leaves the rest of the model’s general abilities intact.

This new approach could help future AI systems track specific objects across time, like a child’s backpack, or localize objects of interest, such as a species of animal in ecological monitoring. It could also aid in the development of AI-driven assistive technologies that help visually impaired users find certain items in a room.

“Ultimately, we want these models to be able to learn from context, just like humans do. If a model can do this well, rather than retraining it for each new task, we could just provide a few examples and it would infer how to perform the task from that context. This is a very powerful ability,” says Jehanzeb Mirza, an MIT postdoc and senior author of a paper on this technique posted to the arXiv preprint server.

Mirza is joined on the paper by co-lead authors Sivan Doveh, a graduate student at Weizmann Institute of Science; and Nimrod Shabtay, a researcher at IBM Research; James Glass, a senior research scientist and the head of the Spoken Language Systems Group in the MIT Computer Science and Artificial Intelligence Laboratory (CSAIL); and others. The work will be presented at the International Conference on Computer Vision (ICCV 2025), held Oct 19–23 in Honolulu, Hawai’i.

An unexpected shortcoming

Researchers have found that large language models (LLMs) can excel at learning from context. If they feed an LLM a few examples of a task, like addition problems, it can learn to answer new addition problems based on the context that has been provided.

A vision-language model (VLM) is essentially an LLM with a visual component connected to it, so the MIT researchers thought it would inherit the LLM’s in-context learning capabilities. But this is not the case.

“The research community has not been able to find a black-and-white answer to this particular problem yet. The bottleneck could arise from the fact that some visual information is lost in the process of merging the two components together, but we just don’t know,” Mirza says.

The researchers set out to improve VLMs abilities to do in-context localization, which involves finding a specific object in a new image. They focused on the data used to retrain existing VLMs for a new task, a process called fine-tuning.

Typical fine-tuning data are gathered from random sources and depict collections of everyday objects. One image might contain cars parked on a street, while another includes a bouquet of flowers.

“There is no real coherence in these data, so the model never learns to recognize the same object in multiple images,” he says.

To fix this problem, the researchers developed a new dataset by curating samples from existing video-tracking data. These data are video clips showing the same object moving through a scene, like a tiger walking across a grassland.

They cut frames from these videos and structured the dataset so each input would consist of multiple images showing the same object in different contexts, with example questions and answers about its location.

“By using multiple images of the same object in different contexts, we encourage the model to consistently localize that object of interest by focusing on the context,” Mirza explains.

Forcing the focus

But the researchers found that VLMs tend to cheat. Instead of answering based on context clues, they will identify the object using knowledge gained during pretraining.

For instance, since the model already learned that an image of a tiger and the label “tiger” are correlated, it could identify the tiger crossing the grassland based on this pretrained knowledge, instead of inferring from context.

To solve this problem, the researchers used pseudo-names rather than actual object category names in the dataset. In this case, they changed the name of the tiger to “Charlie.”

“It took us a while to figure out how to prevent the model from cheating. But we changed the game for the model. The model does not know that ‘Charlie’ can be a tiger, so it is forced to look at the context,” he says.

The researchers also faced challenges in finding the best way to prepare the data. If the frames are too close together, the background would not change enough to provide data diversity.

In the end, finetuning VLMs with this new dataset improved accuracy at personalized localization by about 12% on average. When they included the dataset with pseudo-names, the performance gains reached 21%.

As model size increases, their technique leads to greater performance gains.

In the future, the researchers want to study possible reasons VLMs don’t inherit in-context learning capabilities from their base LLMs. In addition, they plan to explore additional mechanisms to improve the performance of a VLM without the need to retrain it with new data.

“This work reframes few-shot personalized object localization—adapting on the fly to the same object across new scenes—as an instruction-tuning problem and uses video-tracking sequences to teach VLMs to localize based on visual context rather than class priors. It also introduces the first benchmark for this setting with solid gains across open and proprietary VLMs.

“Given the immense significance of quick, instance-specific grounding—often without finetuning—for users of real-world workflows (such as robotics, augmented reality assistants, creative tools, etc.), the practical, data-centric recipe offered by this work can help enhance the widespread adoption of vision-language foundation models,” says Saurav Jha, a postdoc at the Mila-Quebec Artificial Intelligence Institute, who was not involved with this work.

This story is republished courtesy of MIT News (web.mit.edu/newsoffice/), a popular site that covers news about MIT research, innovation and teaching.