Back to list
Osmo Says Its AI Nose Is as Good as a Human’s: How It Builds Training Data
Industry NewsOsmoMachine OlfactionAI Training Data

Osmo Says Its AI Nose Is as Good as a Human’s: How It Builds Training Data

AI startup Osmo says its AI nose is as good as a human’s, according to the headline of Amit Katwala’s MIT Technology Review report. The available excerpt focuses on how the startup builds its training data. In a lab near Newark Airport in New Jersey, around 30 experienced human sniffers are working through odors and recording impressions such as “nutty,” “peachy,” and “marine.” Their notes are cross-referenced with chemical analysis, pairing descriptions of how odors smell to people with information about their chemistry. The excerpt does not explain how Osmo measured its AI nose against human performance or provide results from that comparison.

MIT Technology Review - AI

Key Takeaways

  • Osmo says its AI nose is as good as a human’s. The available excerpt does not include the tests or results behind that comparison.
  • Around 30 experienced human sniffers are recording odor descriptions at a lab near Newark Airport in New Jersey.
  • The team’s notes include terms such as “nutty,” “peachy,” and “marine.” Those descriptions are cross-referenced with chemical analysis to create training data.

In-Depth Analysis

How Osmo pairs human smell descriptions with chemical analysis

The most concrete part of Amit Katwala’s MIT Technology Review excerpt is the account of how Osmo builds its training data. Around 30 experienced human sniffers are working through odors at a lab near Newark Airport. They record what they smell in words, with “nutty,” “peachy,” and “marine” among the examples quoted. Those notes are then cross-referenced with chemical analysis. The two kinds of information answer different questions: the notes describe how an odor was perceived, while the chemical analysis provides information about its chemistry. Their connection is what makes the data relevant to Osmo’s stated goal. The report describes the team as slowly working its way through every odor on Earth, but that wording describes the scope of the work, not a completed inventory. The excerpt gives no count of odors already covered. It also does not say whether sniffers describe each odor independently or how their notes are combined if they use different words. The examples therefore show what the descriptions can look like, not how the team turns multiple observations into a final entry.

What “as good as a human’s” does—and does not—tell us

The headline attributes the performance claim to Osmo. The excerpt, by contrast, shows experienced people helping create the training data. These are two distinct roles for human judgment: people can supply descriptions used in training, and people can serve as a point of comparison in an evaluation. Evidence of the first is not evidence of the second. To interpret the claim, a reader would need to know what the AI and the humans were asked to do. The excerpt does not establish whether the comparison involved describing odors, distinguishing between them, or some other task. Nor does it say who served as the human comparison group or whether those people were the same experienced sniffers who produced the notes. That distinction matters because the headline refers broadly to a human’s sense of smell, while the reporting describes a specific group of experienced assessors. The available text does not connect the two groups through an evaluation or give a result that lets readers judge the comparison for themselves.

Industry Impact

The industry-relevant detail is the work that goes into making human observations usable as AI training data. In Osmo’s case, the excerpt describes neither a collection of chemical analyses on its own nor a list of words on its own. It describes a process for linking the two. That link is central to understanding what the startup is building: the words capture the team’s impressions, while the chemical analysis gives those impressions a chemical reference. The involvement of experienced sniffers is therefore part of the technical story, not just a detail about who works in the lab. Their notes supply the perceptual side of the training data described in the report.

The excerpt also gives readers a way to distinguish a data-building effort from a performance milestone. It shows how Osmo obtains and organizes information relevant to smell, but not how successfully its AI uses that information. The passage ends partway through Osmo’s description of what it calls the world’s first “foundation” something, so the available text does not establish the full wording or substantiate the claim of being first. For the AI industry, the account is informative about the training-data process. It does not establish a benchmark for machine olfaction or show how Osmo compares with other systems.

Frequently Asked Questions

What role do human sniffers play in Osmo’s AI training data?

Around 30 experienced sniffers record their impressions of odors in a New Jersey lab near Newark Airport. The excerpt quotes “nutty,” “peachy,” and “marine” as examples of their notes. Those descriptions are cross-referenced with chemical analysis to create training data.

Does the excerpt show that Osmo’s AI nose matches human performance?

No. The headline reports that Osmo says its AI nose is as good as a human’s. The excerpt does not provide a test procedure, specify what the AI and people were asked to do, or report the results of a comparison. That leaves the claim unverified by the available text, not disproved.

Does the report say Osmo has already cataloged every odor on Earth?

No. The excerpt describes the team as working its way through odors, which indicates an ongoing effort. It does not say the work is finished or specify how many odors the team has covered.

Related News

Japanese Investors Seek $9.4 Billion in Anthropic IPO Shares and Receive $2.98 Billion
Industry News

Japanese Investors Seek $9.4 Billion in Anthropic IPO Shares and Receive $2.98 Billion

Japanese investors sought $9.4 billion in shares in Anthropic’s initial public offering and were ultimately allotted US$2.98 billion, Tech in Asia reported. The report cites a filing for the final allotment. The two figures describe different things: the $9.4 billion is the reported value of the shares Japanese investors sought, while the US$2.98 billion is the value of the shares they were allotted. The allotment was therefore lower than the amount sought. The available report does not identify the investors, explain why the allotment was lower, or give the overall size of the offering. It documents demand and allocation for Japanese investors without establishing how that demand compared with interest from other investors.

Apple Is Hiring to Build an AI Agent That Can Work Across Apps
Industry News

Apple Is Hiring to Build an AI Agent That Can Work Across Apps

Apple is looking for an engineer to lead the creation of its own agentic AI assistant, according to a job listing spotted by Bloomberg and reported by The Verge. The assistant is described as being able to browse the web, work across apps, write code, and call services on a user's behalf. The listing is notable for the range of tasks it names: it describes an assistant that does work for users, rather than establishing how any one capability would function. The available report concerns hiring and the scope of the project. It does not announce a release or say which apps and services the assistant would be able to use.

Meta’s Muse Charm AI Wearable May Cost $300, but Meta Has Not Confirmed the Price
Industry News

Meta’s Muse Charm AI Wearable May Cost $300, but Meta Has Not Confirmed the Price

Meta’s upcoming Muse Charm AI wearable may cost $300, according to a TechCrunch AI article by Lucas Ropek that cites reporting from The Information. Meta did not deny the reporting, saying only that it would “have more to share soon” about the wearable. The distinction is that $300 is a reported possibility, not an official price. Meta’s response indicates that it expects to share more information, but it does not say when that will happen or whether the next update will address pricing. For now, the article establishes that Muse Charm is an upcoming AI wearable and that $300 has been reported as a possible price.