Seeq Expands Machine Learning Support to Democratize Data ...
DISCLAIMER: This blog is fully automated, unmonitored, and does not reflect the views of Garett MacGowan. The ML model may produce content that is offensive to some readers.
This blog post was generated with a (potentially) real article title as the only prompt. A link to the original article is below.
Photo by Jeremy Thomas on Unsplash
Generated: 10/15/2021
Seeq Expands Machine Learning Support to Democratize Data ...and AI
(Photo by Stephen Moulton/Unsplash)
The company behind Questionable Content, a web comic, today announced that it’ll add support for machine-learning (ML) to its content editing tools. Today you don’t have to spend a fortune to try and teach your comic strips to draw or caption stories. You can take a video of a comic strip and upload it to the site with a few lines of text and ask it to “learn” from your video.
The site is built on AI-enabled question-and-answer (Q&A) software, and you can also see in the image below (somewhat) what Q.C. is learning from its users. It draws its visual depictions of the comic from the text that it’s getting from the user, and does so in a way that’s not too distracting. The software is based on a technology called machine reading, which has become a go-to solution for sites like Amazon and Google for automatically translating text from one language to another.
But the thing is, you can’t ask Q.C. to improve the accuracy of its machine reading. “You can use our free version to find the text you want, but the full version is the only that’s going to allow you to teach Q.C. how to improve,” said Seeq, the company behind the site, in an interview with VentureBeat. The company was founded by a set of PhDs and computer scientists from MIT and the NYU Tandon School of Engineering, who are also listed as co-founders of Seeq.
To see the difference, compare the image to the next one: The second image has been edited using a full-featured edition of the software, and I’ve told it to take a look for me at the line, “You’ll think that you’re reading a book that’s been lying on a shelf for years” and make sure that it gets the joke.
(Q.C.)
(The edited image.)
Seeq CEO and co-founder Mark J. Montandon said that one of the coolest features of the platform is the company’s emphasis on privacy. The company’s content analysis software doesn’t run on an Amazon machine, or anywhere else, where someone could access your content data, Montandon said. Its machine reading software is even powered by its own blockchain network, which provides decentralized, open access to the machine reading technology and data.
“Anybody who’s going to do something in machine learning, like an AI or a software engineering firm, has to worry about somebody else taking their data and giving people the best answer for their own product, whether that’s selling your data or selling the service,” he added, noting that Seeq has not made any partnerships or arrangements to sell data or anything like that and has no plans to.
So why has Seeq come together with support for ML today? For one, the technology has seen rapid evolution in the past few years.
“I’ve been in this business for 15 years, and it used to be that you’d say, ‘Here’s this big problem that we have no idea what to do about it,'” Montandon said. “Maybe 10 years ago, you might have thrown a server at it or you may have hired some researchers away from the ivory tower in a university for a couple years and they would do what they could, but not with ML.”
He added, “The technology has changed so dramatically over the last decade or so — the ability to work with very big datasets, to have access, to have the technology to process them, and do all these things, that what’s possible now is way beyond what existed 10 years ago.”
The ability to work with large amounts of data — known as “big data,” or “scale” in other fields — has become important as we’ve witnessed companies like Walmart, Tesla, the U.S. Postal Service, the U.S. government, and Google make sweeping changes to their business operations in the last few years, and AI is starting to power many of their processes. There are already some pretty sophisticated uses of AI — think Watson on Jeopardy, or Alexa — but many of them are pretty boring and mundane. The ML systems of today take things like machine reading and language understanding to new heights.
But the new technologies don’t scale well yet, which means only a select group of companies can build them. You have to have a PhD and hire a PhD-level researcher to work on some pretty interesting problems in that field, something few companies can afford.
So that limits the technology to a limited number of companies and the people who are creating them. The process is similar to today’s closed proprietary technologies that people build, but with limited access.
“It’ll take some time before this technology goes out to the masses and has a lot of the applications in which we would like to provide services, but it’ll be in place sooner than you think,” Montandon predicts. “When there’s a problem that’s not possible to solve, we’ll take a very different approach and that’s what we’re seeing today … We can do stuff with the scale that’s so enormous that would have been impossible even a few years ago.”
Montandon pointed out that the company was built up to provide a new model. Seeq is “not a startup trying to solve a problem or bring in other types of companies, but what we’ve built is an entirely new process for building AI-powered technology in the enterprise,” he said.
Machine reading is an important component of this new process, but so is the blockchain. You should see the blockchain as the platform, and the blockchain can provide the scalability to work with those big datasets, he added.
“Blockchain,” Montandon said, “is a completely different way to think about the concept of information than you have today.”
Blockchain is, simply, a system to record transactions where everyone agrees to the record. It’s the opposite of what we’ve come to think of as information — that it exists only on a server somewhere (unless a user requests access to it) — and that the information is inherently owned by the single point of control. So, while you can “own” information today, because you access it, you don’t own it. You’re in control of where and when. With blockchain, the ability to access data is decentralized instead, meaning there may be multiple points of control.
That’s important because AI requires access to this unstructured data, the kind that can be scattered and hard to find. You can’t easily ask an AI machine if you can see this information. You’ll have to go find it yourself, and that can take days to hours of manually digging through different data servers.
With blockchain, there’s an open access to all information — and every single piece of information in the blockchain already has a unique identifier that provides a path to finding more information, Montandon said.
“The blockchain changes that structure of access to information,” he added. “It’s the same kind of structure that used to be [in the] library or in the world of books, but it’s changed a little bit. There’s now a permanent record of a transaction, so if something changed, we can find the same thing that once required you to go search all over the place and find what’s on someone else’s shelf; we can now find it right away.”
That leads us to another important component of Seeq: The company’s emphasis on decentralization.
“We’re decentralizing the data storage and the data access and the computing for everyone involved, so that you no longer have to trust an organization or an individual to be the provider of any service,” Montandon explained.
“You can go to a centralized place and get whatever data you want, but you no longer have to be a customer of an individual company,” he said.
That’s important because, in the future, AI could provide a lot of critical services. The question now is: Should a large firm like General Motors, with a data set of many hundreds of thousands of transactions, make sure that it can access any one of those transactions from the blockchain?
Or, should a small firm (or individual), like your local car mechanics, be able to access that data and perform their own analysis because they have a smaller dataset?
Montandon acknowledged that the company will have a commercial version of the technology and the product will cost money. There’s also a research version.
“You can now have AI on your phone,” he said in referring to applications using this technology like Amazon’s Alexa or Apple’s Siri. “It’s not hard to imagine a future in which every individual will have their own blockchain.”
There’s only going to be a few large companies that already have the necessary data and can build this platform, but then more will figure it out.
(Photo by Stephen Moulton/Unsplash)
The company behind Questionable Content, a web comic, today announced that it’ll add support for machine-learning (ML) to its content editing tools. Today you don’t have to spend a fortune to try and teach your comic strips to draw or caption stories. You can take a video of a comic strip and upload it to the site with a few lines of text and ask it to “learn” from your video.
The site is built on AI-enabled question-and-answer (Q&A) software, and you can also see in the image below (somewhat) what Q.C. is learning from its users. It draws its visual depictions of the comic from the text that it’s getting from the user, and does so in a way that’s not too distracting. The software is based on a technology called machine reading, which has become a go-to solution for sites like Amazon and Google for automatically translating text from one language to another.
But the thing is, you can’t ask Q.C. to improve the accuracy of its machine reading. “You can use our free version to find the text you want, but the full version is the only that’s going to allow you to teach Q.C. how to improve,” said Seeq, the company behind the site, in an interview with VentureBeat. The company was founded by a set of PhDs and computer scientists from MIT and the NYU Tandon School of Engineering, who are also listed as co-founders of Seeq.
To see the difference, compare the image to the next one: The second image has been edited using a full-featured edition of the software, and I’ve told it to take a look for me at the line, “You’ll think that you’re reading a book that’s been lying on a shelf for years” and make sure that it gets the joke.
(Q.C.)
(The edited image.)
Seeq CEO and co-founder Mark J. Montandon said that one of the coolest features of the platform is the company’s emphasis on privacy. The company’s content analysis software doesn’t run on an Amazon machine, or anywhere else, where someone could access your content data, Montandon said. Its machine reading software is even powered by its own blockchain network, which provides decentralized, open access to the machine reading technology and data.
“Anybody who’s going to do something in machine learning, like an AI or a software engineering firm, has to worry about somebody else taking their data and giving people the best answer for their own product, whether that’s selling your data or selling the service,” he added, noting that Seeq has not made any partnerships or arrangements to sell data or anything like that and has no plans to.
So why has Seeq come together with support for ML today? For one, the technology has seen rapid evolution in the past few years.
“I’ve been in this business for 15 years, and it used to be that you’d say, ‘Here’s this big problem that we have no idea what to do about it,'” Montandon said. “Maybe 10 years ago, you might have thrown a server at it or you may have hired some researchers away from the ivory tower in a university for a couple years and they would do what they could, but not with ML.”
He added, “The technology has changed so dramatically over the last decade or so — the ability to work with very big datasets, to have access, to have the technology to process them, and do all these things, that what’s possible now is way beyond what existed 10 years ago.”
The ability to work with large amounts of data — known as “big data,” or “scale” in other fields — has become important as we’ve witnessed companies like Walmart, Tesla, the U.S. Postal Service, the U.S. government, and Google make sweeping changes to their business operations in the last few years, and AI is starting to power many of their processes. There are already some pretty sophisticated uses of AI — think Watson on Jeopardy, or Alexa — but many of them are pretty boring and mundane. The ML systems of today take things like machine reading and language understanding to new heights.
But the new technologies don’t scale well yet, which means only a select group of companies can build them. You have to have a PhD and hire a PhD-level researcher to work on some pretty interesting problems in that field, something few companies can afford.
So that limits the technology to a limited number of companies and the people who are creating them. The process is similar to today’s closed proprietary technologies that people build, but with limited access.
“It’ll take some time before this technology goes out to the masses and has a lot of the applications in which we would like to provide services, but it’ll be in place sooner than you think,” Montandon predicts. “When there’s a problem that’s not possible to solve, we’ll take a very different approach and that’s what we’re seeing today … We can do stuff with the scale that’s so enormous that would have been impossible even a few years ago.”
Montandon pointed out that the company was built up to provide a new model. Seeq is “not a startup trying to solve a problem or bring in other types of companies, but what we’ve built is an entirely new process for building AI-powered technology in the enterprise,” he said.
Machine reading is an important component of this new process, but so is the blockchain. You should see the blockchain as the platform, and the blockchain can provide the scalability to work with those big datasets, he added.
“Blockchain,” Montandon said, “is a completely different way to think about the concept of information than you have today.”
Blockchain is, simply, a system to record transactions where everyone agrees to the record. It’s the opposite of what we’ve come to think of as information — that it exists only on a server somewhere (unless a user requests access to it) — and that the information is inherently owned by the single point of control. So, while you can “own” information today, because you access it, you don’t own it. You’re in control of where and when. With blockchain, the ability to access data is decentralized instead, meaning there may be multiple points of control.
That’s important because AI requires access to this unstructured data, the kind that can be scattered and hard to find. You can’t easily ask an AI machine if you can see this information. You’ll have to go find it yourself, and that can take days to hours of manually digging through different data servers.
With blockchain, there’s an open access to all information — and every single piece of information in the blockchain already has a unique identifier that provides a path to finding more information, Montandon said.
“The blockchain changes that structure of access to information,” he added. “It’s the same kind of structure that used to be [in the] library or in the world of books, but it’s changed a little bit. There’s now a permanent record of a transaction, so if something changed, we can find the same thing that once required you to go search all over the place and find what’s on someone else’s shelf; we can now find it right away.”
That leads us to another important component of Seeq: The company’s emphasis on decentralization.
“We’re decentralizing the data storage and the data access and the computing for everyone involved, so that you no longer have to trust an organization or an individual to be the provider of any service,” Montandon explained.
“You can go to a centralized place and get whatever data you want, but you no longer have to be a customer of an individual company,” he said.
That’s important because, in the future, AI could provide a lot of critical services. The question now is: Should a large firm like General Motors, with a data set of many hundreds of thousands of transactions, make sure that it can access any one of those transactions from the blockchain?
Or, should a small firm (or individual), like your local car mechanics, be able to access that data and perform their own analysis because they have a smaller dataset?
Montandon acknowledged that the company will have a commercial version of the technology and the product will cost money. There’s also a research version.
“You can now have AI on your phone,” he said in referring to applications using this technology like Amazon’s Alexa or Apple’s Siri. “It’s not hard to imagine a future in which every individual will have their own blockchain.”
There’s only going to be a few large companies that already have the necessary data and can build this platform, but then more will figure it out.