Microsoft has introduced AI models that it trained internally and says it will begin using them in some products. This announcement may represent an effort to move away from dependence on OpenAI, despite Microsoft's substantial investment in that company. It comes more than a year after insider reports revealed that Microsoft was beginning work on its own foundational models.

A post on the Microsoft AI blog describes two models. MAI-Voice-1 is a natural speech-generation model meant to deliver "high-fidelity, expressive audio across both single and multi-speaker scenarios." The idea is that voice will be one of the main ways users interact with AI tools in the future, though we haven't really seen that come to fruition so far.

The second model is called MAI-1-preview, and it's a foundational large language model specifically trained to drive Copilot, Microsoft's AI chatbot tool. It was trained on around 15,000 Nvidia H100 GPUs, and runs inference on a single GPU. As reported last year, this model is significantly larger than the models seen in Microsoft's earlier experiments, which focused on smaller models meant to run locally, like Phi-3 .

Read full article

Comments

Microsoft has introduced AI models that it trained internally and says it will begin using them in some products. This announcement may represent an effort to move away from dependence on OpenAI, despite Microsoft's substantial investment in that company. It comes more than a year after insider reports revealed that Microsoft was beginning work on its own foundational models.

A post on the Microsoft AI blog describes two models. MAI-Voice-1 is a natural speech-generation model meant to deliver "high-fidelity, expressive audio across both single and multi-speaker scenarios." The idea is that voice will be one of the main ways users interact with AI tools in the future, though we haven't really seen that come to fruition so far.

The second model is called MAI-1-preview, and it's a foundational large language model specifically trained to drive Copilot, Microsoft's AI chatbot tool. It was trained on around 15,000 Nvidia H100 GPUs, and runs inference on a single GPU. As reported last year, this model is significantly larger than the models seen in Microsoft's earlier experiments, which focused on smaller models meant to run locally, like Phi-3 .

Read full article

Comments

Microsoft has introduced AI models that it trained internally and says it will begin using them in some products. This announcement may represent an effort to move away from dependence on OpenAI, despite Microsoft's substantial investment in that company. It comes more than a year after insider reports revealed that Microsoft was beginning work on its own foundational models.

A post on the Microsoft AI blog describes two models. MAI-Voice-1 is a natural speech-generation model meant to deliver "high-fidelity, expressive audio across both single and multi-speaker scenarios." The idea is that voice will be one of the main ways users interact with AI tools in the future, though we haven't really seen that come to fruition so far.

The second model is called MAI-1-preview, and it's a foundational large language model specifically trained to drive Copilot, Microsoft's AI chatbot tool. It was trained on around 15,000 Nvidia H100 GPUs, and runs inference on a single GPU. As reported last year, this model is significantly larger than the models seen in Microsoft's earlier experiments, which focused on smaller models meant to run locally, like Phi-3 .

Read full article

Comments

Microsoft's fifth major iteration of Windows 11 is nearing its release to the general public—the Windows Insider team announced today that Windows 11 25H2 was being put into its Release Preview Channel, the final stop for most updates before they become available to everyone. That's around two months after the first Windows builds with the 25H2 label were released to the other preview channels.

Putting a new yearly Windows update in the Release Preview channel is analogous to the "release to manufacturing" (RTM) phase of years past, back when updates shipped on physical media that needed to be manufactured. Build numbers for this version of Windows start with 26200, rather than 24H2's 26100.

The 25H2 update doesn't do a lot in and of itself, other than reset the clock for Microsoft's security updates (each yearly release gets two years of security patches). Microsoft says that last year's 24H2 update and this year's 25H2 update "use a shared servicing branch," which mostly means that there aren't big under-the-hood differences between the two. Installing the 25H2 update on a PC may enable some features on your 24H2 PC that had already been installed but had been disabled by default.

Read full article

Comments

Microsoft's fifth major iteration of Windows 11 is nearing its release to the general public—the Windows Insider team announced today that Windows 11 25H2 was being put into its Release Preview Channel, the final stop for most updates before they become available to everyone. That's around two months after the first Windows builds with the 25H2 label were released to the other preview channels.

Putting a new yearly Windows update in the Release Preview channel is analogous to the "release to manufacturing" (RTM) phase of years past, back when updates shipped on physical media that needed to be manufactured. Build numbers for this version of Windows start with 26200, rather than 24H2's 26100.

The 25H2 update doesn't do a lot in and of itself, other than reset the clock for Microsoft's security updates (each yearly release gets two years of security patches). Microsoft says that last year's 24H2 update and this year's 25H2 update "use a shared servicing branch," which mostly means that there aren't big under-the-hood differences between the two. Installing the 25H2 update on a PC may enable some features on your 24H2 PC that had already been installed but had been disabled by default.

Read full article

Comments

Microsoft's fifth major iteration of Windows 11 is nearing its release to the general public—the Windows Insider team announced today that Windows 11 25H2 was being put into its Release Preview Channel, the final stop for most updates before they become available to everyone. That's around two months after the first Windows builds with the 25H2 label were released to the other preview channels.

Putting a new yearly Windows update in the Release Preview channel is analogous to the "release to manufacturing" (RTM) phase of years past, back when updates shipped on physical media that needed to be manufactured. Build numbers for this version of Windows start with 26200, rather than 24H2's 26100.

The 25H2 update doesn't do a lot in and of itself, other than reset the clock for Microsoft's security updates (each yearly release gets two years of security patches). Microsoft says that last year's 24H2 update and this year's 25H2 update "use a shared servicing branch," which mostly means that there aren't big under-the-hood differences between the two. Installing the 25H2 update on a PC may enable some features on your 24H2 PC that had already been installed but had been disabled by default.

Read full article

Comments

At the beginning of the month, Tesla was found partly liable in a wrongful death lawsuit involving the death of a pedestrian in Florida in 2019. The automaker— which could have settled the case for far less —claimed that it did not have the fatal crash's data. That's until a hacker was able to recover it from the crashed car, according to a report in The Washington Post .

In the past , Tesla has been famously quick to offer up customer data stored on its servers to rebut claims made against the company. But in this case, the company said it had nothing. Specifically, the lawyers for the family wanted what's known as the "collision snapshot," data captured by the car's cameras and other sensors in the seconds leading up to and after the crash.

According to the trial, moments after the collision snapshot was uploaded to Tesla's servers, the local copy on the car was marked for deletion. Then, "someone at Tesla probably took 'affirmative action to delete' the copy of the data on the company’s central database," according to the Post.

Read full article

Comments

At the beginning of the month, Tesla was found partly liable in a wrongful death lawsuit involving the death of a pedestrian in Florida in 2019. The automaker— which could have settled the case for far less —claimed that it did not have the fatal crash's data. That's until a hacker was able to recover it from the crashed car, according to a report in The Washington Post .

In the past , Tesla has been famously quick to offer up customer data stored on its servers to rebut claims made against the company. But in this case, the company said it had nothing. Specifically, the lawyers for the family wanted what's known as the "collision snapshot," data captured by the car's cameras and other sensors in the seconds leading up to and after the crash.

According to the trial, moments after the collision snapshot was uploaded to Tesla's servers, the local copy on the car was marked for deletion. Then, "someone at Tesla probably took 'affirmative action to delete' the copy of the data on the company’s central database," according to the Post.

Read full article

Comments

At the beginning of the month, Tesla was found partly liable in a wrongful death lawsuit involving the death of a pedestrian in Florida in 2019. The automaker— which could have settled the case for far less —claimed that it did not have the fatal crash's data. That's until a hacker was able to recover it from the crashed car, according to a report in The Washington Post .

In the past , Tesla has been famously quick to offer up customer data stored on its servers to rebut claims made against the company. But in this case, the company said it had nothing. Specifically, the lawyers for the family wanted what's known as the "collision snapshot," data captured by the car's cameras and other sensors in the seconds leading up to and after the crash.

According to the trial, moments after the collision snapshot was uploaded to Tesla's servers, the local copy on the car was marked for deletion. Then, "someone at Tesla probably took 'affirmative action to delete' the copy of the data on the company’s central database," according to the Post.

Read full article

Comments