FreeGameGuide spoke with Marcio Paulino, Technology Leader at Intel Brazil, who revealed the development of security in the use of artificial intelligence
At the CloudOps Summit by RealCloud event, which took place this Thursday (07), the Canaltech spoke with Marcio Paulino, Technology Leader at Intel Brazil, about artificial intelligence solutions that are making their way to the end consumer, as well as the security aspects of enterprise technology that are making their way to the consumer public.
- How Intel’s NPU and AI can help you in your everyday life
- Intel announces new features to “democratize” the use of artificial intelligence
We asked what the outlook was Intel for the artificial intelligence (AI) market and the improvements that this will bring to the common consumer, thinking about the improvements that exist in the business market. In addition to highlighting the democratization of AI, Marcio said the global estimate for the AI market is $7 billion for CPU alone, not taking graphics chips into account.
“Intel is putting components to accelerate CPU performance for AI workloads. One example is that we put into Sapphire Rapids, fourth-generation Xeon, an accelerator called Intel AMX, which gives the processor the ability to perform matrix calculations , matrix multiplication, which is what a neural network does. This is what we are bringing to democratize the use of artificial intelligence and make the user spend less, because specific hardware costs a lot and the idea of Intel is that, for every CPU launched, we will put a capability to run artificial intelligence, “the technology leader emphasized.
🛒Buy the Intel Core i9-13900KF at the best price!
🛒Buy the Intel Arc A750 at the best price!
Secure AI with Intel technologies
During the CloudOps Summit, many security proposals were shown for companies, while consumers, who will have increasingly access to artificial intelligence, will not be able to count on security at an enterprise-level price. So, how will AI help the end consumer in terms of security? The Intel Brazil executive stressed that the company continues to use several technologies to always protect the user:
“We have a technology called Intel TDX and Intel SGX in our processors. When we train a neural network, we need data. This data is cleaned when it is running on the CPU and this is the ‘best time’ for a hacker to attack, because data is clean at run time. If it arrives encrypted at run time, it cannot be encrypted. Intel introduced these features exactly to protect and ensure that customers comply with AI training and care about data used for training.”
OR IntelTDX AND Intel SGX they are Intel Trust Domain Extensions and Intel Software Guard Extensions respectively. While Intel TDX creates a secure space to run without external access, Intel SGX performs encryption at a granular level in the hardware. Paulino also stressed that “Intel is offering it in an economical way and the user will now be able to have these technologies integrated.”
The Blue Team specialist also highlighted performance and safety as pillars of this movement. Therefore, the neural engine will be used to optimize processing and also improve security, highlighting that the NPU of the Ultra Core it will have the same level of protection as the Xeon.
“We will launch Core Ultra which has a neural engine that helps learn user behavior and anticipate it using neural processing. […] This NPU will help precisely at this point, having a personalized assistant and, in terms of security, Intel puts the same execution protection on the Core Ultra that it has on the Xeon, to protect this model of yours that is learning its craft – daily life.” , underlined the head of technology of Intel Brazil.
Arc and generative AI
As a final topic, we asked Marcio Paulino Generative AI, which has become very popular with ChatGPT, and how Intel technology will help people. He highlighted how the vast majority of AI models start from a notebook and, therefore, having a network capable of supporting a good number of parameters can help in the development of more advanced technologies.
“Generative AI, or AI in general, needs a good platform, a good CPU, a good GPU, in this case ARC, and also a good NPU, mainly for NPU inference and creation of GPUs. […] A data scientist, when assembling a neural network, assembles it on a notebook. When he has an efficient CPU, an efficient GPU and an NPU, he will be able to extend the neural network a little further at creation time, before migrating to a server, which costs more. So he is much more decisive when he climbs, since he has tested these characteristics in his notebook,” Marcio said.
Finally, he highlighted how the new Intel technology is ready to start working with generative artificial intelligence and bring it to the public, regardless of their level of knowledge on the subject.
“Speaking of generative artificial intelligence, a good model is between 7 and 12 billion parameters: they are like the connections of neurons. ChatGPT has 175 billion parameters, but with 2 and 13 billion parameters, our assistant will “fly”, and therefore this hardware with this capacity will be able to satisfy this demand that users will begin to have in the response to generative artificial intelligence”, concluded the technology leader of Intel Brazil.
Trends on Canaltech:
- A meteor 40 times faster than the plane was recorded in SC and SP
- The drone captures the sensational maneuver of a giant ship in the port of Santos
- The 50 funniest Google Assistant jokes
- Review The world after us | In the midst of chaos, who can we trust?
- The 15 best Android apps of 2021
- What does the name Yu Yu Hakusho mean?
Source: Terra
Rose James is a Gossipify movie and series reviewer known for her in-depth analysis and unique perspective on the latest releases. With a background in film studies, she provides engaging and informative reviews, and keeps readers up to date with industry trends and emerging talents.
