Obo’s creators know what everyone is thinking about AI. “Is AI going to fool us all?” The company asks in a recent advertisement. “Are we going to forget how to think for ourselves?”
Obo founders feel that both those questions are answered NoAnd their startup is to prove it.
The AI Shiksha Manch, which was launched this month, uses AI to crafts the “course” about any subject that attacks the user’s fancy. It comes from the creators of the anchor, a DIY mini-podcast creation platform, acquired by Spotify in $ 150 million. Nir Zicherman, OBOE CEO and Cofounder, ended running the audiobook vertical of Spotify, while Mike Mignano, an OBOE COFOUNing also, drove the podcast team of Spotify.
Zicherman wants to “democratization of access to a great learning experience” on cheap with AI, he tells The verge This is an sublime claim, especially given how many products have been branded as already AI-operated learning tools-and how the AI products continue to have large-scale hallucinations.
Currently, people turn to learn chat, Google, YouTube, Wikipedia and other internet platforms, argument of Zicherman. In this process users need to “cut it together” and “is a very linear journey that is a shape-fit-everyone,” they say. Meanwhile, Obo “streamlined the information for a single destination, which you really need to go to learn effectively.”
Obo’s website has an familiar chatbot that invites users to invite users to learn what they want to learn. Unlike chat, however, OBOE will not have a back and forth with users. Instead, the obo’s response to a sign will be its signature “course” about the subject using traditional educational formats, such as a long written text that appears similar to a small bullet list of an introductory chapter or “key takeaairs” in a long-term written text.
One can produce “courses” on the origin of the forms of artificial intelligence or intelligence in nature, both depicted in Obo’s advertisement.
Or, you can be like me and choose uncontrolled, opaque subject of concrete manufacturing and its environmental effects, about which I know nothing about.
This produced a “deep dive” essay to me, which was labeled as a 10 -minute reed. “Imagine fresh concrete in your palm,” the section began. The text was broken by a header such as the “environmental footprint” and was combined with tables that listed some types of “SCM”. What is SCM? I hit the command + F on the page for the definition of “SCM”, but it was not. Pictures of Wikimedia Commons appeared in every few paragraphs. After “deep dive”, I was able to click through information packed in various formats like an FAQ and podcast.
When I felt a cursory understanding of concrete construction, I went away, Obo has not solved the biggest problem in AI for me yet: I did not know that any information was accurate. The details are confirmed to the learner. For example, for example, OBOE does not include links to the original source material, for example, “set granular materials such as sand, gravel, or crushed stones that form a majority of concrete volume (usually 60–75%).” It came out Correct According to the American Cement Association, but I had to detect my possible source to verify the statistical.
Obo is working on connecting the road down, but it is not available today. Georousman wrote in an email comment, “Quotes and other means of accessing additional online resources are therefore we are actively working and expect to add to the stage in the coming months.” Ruckus,
Other platforms give you quotes, when you ask for them. When I turned to Google’s Gemini to know about the concrete and included in a request for my quick quotes, I was given A Link to a 2018 paper In the journal Nature stability It was calculated that about 10 percent of the world around the world eat global concrete production for the use of industrial water. useful! Perhaps quotes generally give me a wrong meaning of safety that the information is accurate and some AI Slope is not being drawn from the website. But at least the option is there if I care about going in search of it.
Obo does not train his own fundamental AI model, says odorman, and chooses models from other companies instead-he will not name them-which are adapted to some tasks, such as speech-text synthesizers or large language models. Zicherman says that OBOE addresses some llms to correct the output of other llms, with the work of potential hallucinations material. In other words, LLMs are investigating other LLMs.
“For example, you may have a single model output that assumes that it is a certain set of facts that should be included in a course, and then another model from a separate provider who has been trained in a different dataset in a different way, in a way, a way, which identifies and helps to reduce the possibility of malaise.”
Does it fail, yet there is no facility that allows users to flagged wrong material within their output, Therefore, for now, users need to report impurities directly to the company. Littleman says that they have a process to include “pipeline” improvement “improvement” “.
If you do not want to read, Obo also produces podcast. A pair of Chipper, AI-Janit “Host” broke the environmental footprint of concrete in a connivance manner as NPR Planetary currency For me.
If AI-Janit Podcast looks similar to Google’s notebook, it is. The difference is that OBOE users do not need to upload materials to generate podcasts, but still, the material can upload. But this freedom comes to a cost – I am confident that the podcast “host” of the notebook is more than the relatively more obo because I know the information Comes from my documents – At least, it should be. Meanwhile, Obo’s podcasts are made up of anonymous models without quotes, so I could not bring myself to trust what I heard.
Zicherman says that OBOE will continue to improve because more users feed more “courses” and learning styles and users conflicts back into the model. He likes obo from a human tutor who learns how the student learns over time. “OBOE is not just a product that gives you a great personal experience, it is also a product that gets better that you use it more, the way you spent more time with them as a human tutor and they were able to understand more and more about how you learn effectively.”
For now, I still prefer to hear directly from human experts – even though I need to work hard to identify them on the huge ocean of the Internet, the AI slope flooded rapidly.
0 Information


