Page 54 - SAMENA Trends - September 2023

ARTICLE SAMENA TRENDS

video data. With this, it takes only 5 seconds To meet enterprises' data Generative AI is a rapidly developing
to generate a 3D digital human model from technology with the potential to impact
a photo, with support for personalized security requirements for society significantly, both positively and
facial adjustments. The "digital human AI models, Huawei Cloud negatively. As a result, governments and
video production" service allows users to provides multiple deploy- international organizations are developing
create live broadcasts using text, voice, or policies to ensure that Gen AI is used safely
video input. ment modes, including and ethically.
public cloud, dedicated
In the Scientific Computing field, the For example, the European Union's Artificial
Huawei Cloud Pangu Drug Molecule Model zone (public cloud), and Intelligence Act, which is currently being
can generate millions of new molecular hybrid cloud. In this way, drafted, is expected to be one of the
structures and analyze their drug properties most comprehensive AI regulations in the
to screen potential molecular structures as enterprises can enjoy world. The act will set out requirements
drugs. This can reduce the costs of trial the continuous supply of for companies that develop, deploy, and
and error, accelerating the discovery of lead compute resources from use AI systems, including transparency,
compounds from several years to just one accountability, and risk assessment.
month. the public cloud while
keeping data within their The regulatory landscape for AI is
The unprecedented generative AI continually evolving and the level of
opportunity own organizations to build awareness and compliance varies from
Artificial Intelligence (AI) has become an their own AI models. one company to another. Therefore,
integral part of today's world, transforming organizations need to stay updated on
how we live, work, and interact. AI is outperforming peer models in terms regional and global policies surrounding AI,
powering innovations across various of understanding and generation. The as compliance can be complex and rapidly
industries, shifting from traditional AI decoupled, hierarchical architecture allows changing.
models focusing on classification and the Pangu models to be quickly adapted
prediction to AI-Generated Content (AIGC) to a wide range of downstream tasks. In addition to being aware of regional
or Generative AI. According to Gartner, AIGC Customers can load independent datasets and global policies, companies must also
will account for 10% of all data generated to train their own models. They can also be mindful of the ethical implications of
by 2025. Additionally, 30% of outbound choose to upgrade foundation models or using Gen AI. This includes issues such
marketing messages will be generated just upgrade capability sets. as bias, transparency, and accountability.
by AI by 2025, a significant increase from Companies need to ensure that their AI
less than 2% in 2022. We can also expect To meet enterprises' data security systems are fair, unbiased, and transparent
a major blockbuster with 90% of the film requirements for AI models, Huawei Cloud about how they work. They also need to
generated by AI in 2025, from 0% of such in provides multiple deployment modes, be accountable for the decisions made by
2022. I firmly believe that using AI in various including public cloud, dedicated zone their AI systems.
industries will be the next big tipping (public cloud), and hybrid cloud. In this
point, ushering in a new era of ubiquitous way, enterprises can enjoy the continuous By being aware of the regional and
connectivity and intelligence and driving supply of compute resources from the global policies and ethical implications
innovation and productivity across sectors. public cloud while keeping data within their surrounding the use of Gen AI, companies
own organizations to build their own AI can help to ensure that this technology is
Huawei Cloud Pangu streamlines AI models. used safely and responsibly.
adoption for industry Our models are industry-tailored with
Implementing AI into business operations specific knowledge and experiences, such Generative AI will completely transform
requires overcoming a few challenges, as our own business experience over the creative industries
such as difficult data acquisition, difficult past 30 years, know-how in more than ten A Generative AI model is a machine
industry knowledge distillation, and difficult industries, and those from our customers learning-based technology that learns
knowledge computing that enterprises and partners. In addition to an ocean of from large amounts of data and generates
need to address. With conventional AI general knowledge, Pangu models have new content. Compared with traditional AI
model development, a considerable been pre-trained with open datasets from models, this technology is more flexible
amount of manpower, customization, and more than ten industries, including finance, and creative and can produce more diverse
sample training are required for every government, meteorology, healthcare, results. It can be used in natural language
specific scenario and often take a long time Internet, education, automotive, and retail. processing, image generation, music
to develop. The data volume of each industry exceeds creation and other fields, bringing new
50 billion tokens. experiences and possibilities to people.
Huawei Cloud Pangu Models are trained Navigating the policy and ethical landscape In the field of natural language processing,
with hundreds of billions of parameters, of Generative AI Generative AI models can be used for text

54 SEPTEMBER 2023