Intel's Norberto Mateos: AI must be transparent so that the user can decide

Monday 18 Sep 2023

Intel is going to build a robust open-source AI research and development environment.

Norberto Mateos
Norberto Mateos, director general de Intel Iberia


Everything points to 2023 as the year in which generative artificial intelligence arrived to change everything. This advance, from multiple proposals, has generated an intense debate for or against, with thousands of users generating images based on a simple phrase. Intel, like many other technological giants, is also doing its bit with AI that generates 3D content, such as its model (LDM3D), although these are not the only advances in disruptive matters in which the company's laboratories are immersed.

The 700 employees that make up Intel Labs are divided between engineers, computer scientists, physicists, neuroscientists, and anthropologists renowned in their respective specialties in order to cover up to 30 technical disciplines in their multiple research projects. Recently from Labs they have launched LDM3D to create three-dimensional spaces effortlessly, and Tunnel Falls, their most advanced chip that aims to boost qubit research for quantum computing.

Heading into the latter part of the year, Intel Labs is focused on creating breakthroughs in neuromorphic computing, artificial intelligence, and security. Another area of study for these teams is the detection of deepfakes. Intel has created FakeCatcher and assures this media that in a few months it will be able to present new developments in this area.

Norberto Mateos, EMEA Partner Sales Director, explains to Ahram the objectives set by Intel Labs in its multiple research projects. "I believe it is essential to establish measures that guarantee the responsible use of AI through regulations that provide confidence and ensure human oversight, transparency, security, safety and reliability, personal privacy, and fairness and inclusion," he explains.

Intel has recently introduced Latent Diffusion Model for 3D (LDM3D) to generate 3D images, what can we expect from this technology?

LDM3D represents a revolutionary breakthrough for the industry as the first model that uses diffusion to generate a depth map, enabling the creation of 3D images with 360-degree panoramic views. This innovation has the potential to completely transform content production, metaverse applications and digital experiences.

The advent of this model sets the stage for further advances in multiview generative artificial intelligence and computer vision. Intel will continue to explore its use to empower human capabilities and build a robust open-source AI research and development environment, with the goal of democratizing access to this technology. In line with Intel's commitment to an open ecosystem in the field of artificial intelligence, LDM3D is being presented as an open-source project through HuggingFace. This will enable AI researchers and practitioners to further enhance this system and tailor it to custom applications, thus fostering collaboration and continuous improvement.

As an artificial intelligence systems development lab, how do you assess the European artificial intelligence bill?

We have all witnessed the rise of generative artificial intelligence applications that have permeated our daily lives in recent times. These tools have proven to be incredibly compelling, so much so that they have come to confuse us, making it difficult to distinguish between what is AI-generated and what is real. It is therefore essential that AI systems meet certain standards of transparency so that users can make informed decisions. We are confident that the European AI bill will allow the necessary limits to be set to bring out the full potential of this highly powerful technology.

Companies must play a key role in controlling their developments. At Intel, we have established five fundamental pillars for responsible AI. First, a review process has been established in which Intel's Multidisciplinary Advisory Board on Responsible AI reviews the entire lifecycle of an AI project and conducts an ethical impact assessment based on six key areas: human rights, human oversight, explainability, safety, security and reliability, personal privacy, and equity and inclusion. Second, Intel ensures that AI practitioners and their respective technologies are equitable and inclusive. We also recognize the need to include ethics as a fundamental part of any AI education program.

Furthermore, in terms of privacy and security, Intel strives to secure AI and maintaining data integrity, privacy and accuracy is at the core of Intel's research and development efforts in security and holistically, with innovations across hardware and software to enable the ecosystem to build trusted AI.

The growing debate about the risks of AI has now cast an image of this technology as highly dangerous, more so in fields such as medicine. For its part, Intel Labs is training surgical robots to perform sutures.

As motion control technologies have advanced, surgical assisting robots have improved in accuracy. These robots help surgeons achieve higher levels of speed and precision, enabling them to perform complex operations and can even perform tasks autonomously, allowing surgeons to monitor procedures from a console.

Robotics also plays a key role in surgeon education. Simulation platforms use artificial intelligence and virtual reality to provide surgical training in robotics. Within a virtual environment, surgeons can practice procedures and hone their skills using robotic controls.

The opportunities and benefits provided by these innovations are evident, but there is still a lack of limitations that are perceived as a threat by a high percentage of society. Therefore, I believe it is essential to establish measures that ensure the responsible use of AI through regulations that provide confidence and ensure human oversight, transparency, safety, security and reliability, personal privacy, and fairness and inclusion.

In addition, quantum computing is another of the hopes for the technological future. In this area, where Intel has presented Tunnel Falls, what does this new chip offer in the advancement of quantum computing?

Intel's quantum computing vision focuses on taking quantum out of the lab and into commercial reality. Currently, academic institutions lack large-scale fabrication facilities like those offered by Intel. With Tunnel Falls, researchers can start their projects immediately, rather than having to try to create their own devices.

The 12-qubit device is fabricated on 300-millimeter wafers and leverages Intel's innovative industrial transistor manufacturing capabilities. This includes the use of advanced technologies such as extreme ultraviolet (EUV) lithography and gate and contact processing techniques. These technologies enable greater precision and efficiency in transistor manufacturing, resulting in improved 12-qubit device performance and functionality.

We believe this is the right approach. If we build silicon spin qubits on silicon wafers and develop a qubit technology that looks like a transistor, we can follow Moore's Law of acceleration to build a large-scale system. And although it has taken us several years to move a lab prototype to a manufacturing process, we are getting there.


Search Keywords:
Short link: