Prosperity Through Data Infrastructure

Arguments

Legacy systems play a significant role in digitalization

Supporting facts:

• Slovenia is considered as super digital from Swiss perspective

• Telecommunication companies have large legacy systems

Topics: legacy systems, digitalization, innovation

---

Report

The role of legacy systems in the process of digitalisation has been the subject of much discussion and debate. According to Swiss perspective, Slovenia is considered to be highly digital. However, it has been found that telecommunication companies in Slovenia rely heavily on large legacy systems.

This finding suggests that legacy systems continue to play a significant role in the digitalisation process. However, there are arguments suggesting that legacy systems present challenges in the journey of digitalisation. One viewpoint is that these legacy systems hinder progress towards SDG 9, which focuses on industry, innovation, and infrastructure.

This indicates that legacy systems create obstacles to achieving the goals set by SDG 9. While the presence of large legacy systems in Slovenian telecommunication companies supports the argument of their significance in digitalisation, specific supporting evidence was not provided. Therefore, further research and evidence are needed to strengthen and deepen our understanding of the relationship between legacy systems and digitalisation.

In conclusion, the role of legacy systems in digitalisation remains a topic of discussion. On one hand, it is recognised that legacy systems still have a significant impact on the digitalisation efforts of Slovenian telecommunication companies. On the other hand, there are concerns that legacy systems may pose challenges to achieving the objectives of SDG 9.

However, additional specific evidence is needed to further support these arguments and provide a comprehensive understanding of the dynamics between legacy systems and digitalisation.

Arguments

Governments need to collaborate for the development in AI and data infrastructure

Supporting facts:

• Governments cannot do a lot of things on their own

• Public-private collaboration is crucial for addressing challenges

Topics: Governmental role, Collaboration, AI, Data infrastructure

---

Legislation should be predictable, understandable and adaptable as per rapid technological advancements

Supporting facts:

• Companies need to know how future legislation will affect their investments

• The quick development of technology requires adaptable legislation

Topics: Legislation, Technological Developments, AI, Data infrastructure

---

Involvement of various sectors including universities, municipalities and local communities is important

Supporting facts:

• Solutions for digital transition are intended for people living in municipalities

• Municipalities are asked to share data with private organizations for solution development

Topics: Public-private partnerships, Universities, Municipalities, Local communities

---

Interoperability is key for global inclusivity with AI systems

Supporting facts:

• Without interoperability, global inclusive systems can't be established.

• Interoperability refers to the capability of different systems to connect and effectively operate together

Topics: Interoperability, Artificial Intelligence, Inclusivity

---

There is a need to connect the 2.6-2.7 billion people who are currently unconnected

Supporting facts:

• Current AI systems and data are biased as they exclude those who are not connected.

• Inclusive AI systems can't be established without including everyone.

Topics: Digital connectivity, Internet access, Inequality

---

Ensuring interoperability in technology development is a collective responsibility

Supporting facts:

• Governments and industries are instrumental in ensuring interoperable solutions

• International organizations and platforms like the World Economic Forum also play a major role

• The International Telecommunication Union is often responsible for developing relevant standards

Topics: Technology, Interoperability, Standards

---

Legacy systems can pose challenges in digitalization, but you must work with what you have

Supporting facts:

• Slovenia had to work with existing systems and technologies due to its history and past dependencies

Topics: Digitalization, Legacy Systems

---

Emilia believes we need to invest both in technology and in people

Supporting facts:

• Review of the budget of her ministry revealed that more was invested in training and education programs for digital skills

• Simultaneous investment in technology (R&D) and people is necessary

Topics: Investment, Technology, People, Skills

---

Report

The analysis underscores the significance of collaboration between governments and the private sector in the development of artificial intelligence (AI) and data infrastructure. It emphasises that governments, on their own, are unable to address the challenges associated with AI and data infrastructure, highlighting the need for public-private cooperation.

This collaborative approach is considered crucial for effectively tackling these challenges. Another key argument put forth in the analysis is the need for legislation that is predictable, understandable, and adaptable in the face of rapid technological advancements. It is argued that companies need to have a clear understanding of how future legislation will impact their investments to make informed decisions.

Furthermore, the fast-paced development of technology necessitates legislation that can adapt to changing circumstances and remain relevant. The involvement of various sectors, including universities, municipalities, and local communities, is identified as important in addressing the digital transition. Solutions for the transition are intended to benefit people living in municipalities, and as such, it is important for these municipalities to share data with private organizations for the development of effective solutions.

Interoperability is highlighted as a key factor in achieving global inclusivity with AI systems. It is pointed out that without interoperability, it is not possible to establish inclusive systems on a global scale. Interoperability refers to the ability of different systems to connect and operate together seamlessly.

The analysis also raises concerns about the current state of digital connectivity and internet access. It is argued that existing AI systems and data are biased, as they exclude individuals who are not connected to the internet. Inclusive AI systems cannot be established without including everyone, which includes the 2.6-2.7 billion people who are currently unconnected.

Therefore, there is a need to connect these individuals to ensure that AI systems are truly inclusive. Furthermore, the analysis emphasises the importance of starting AI development with unbiased and inclusive data. It is stated that generative AI relies on the data available, and if 2.7 billion people are excluded from this data, they are also excluded from the results and benefits of AI.

Therefore, it is argued that AI should be developed with data that is unbiased and includes everyone to avoid exclusion. Additionally, the analysis highlights the collective responsibility of governments, industries, and major international organizations in ensuring interoperability in technology development.

The International Telecommunication Union is specifically mentioned as having a key role in the development of relevant standards. The challenges posed by legacy systems in digitalisation are acknowledged. It is noted that Slovenia had to work with existing systems and technologies due to its history and past dependencies.

Despite the challenges, the analysis suggests that successfully digitalising requires creative solutions, even when faced with older and potentially outdated systems. Slovenia's example is cited as an illustration of utilising what is available and making the best of it. Lastly, the analysis emphasises the need for investment in both technology and people.

The importance of investing in training and education programmes for digital skills is highlighted, along with simultaneous investment in technology research and development. The analysis contends that investment should not be an "either-or" scenario but rather a combination of investing in both technology and people.

In conclusion, the analysis underscores the importance of collaboration between governments and the private sector in the development of AI and data infrastructure. It also stresses the need for predictable and adaptable legislation, the involvement of various sectors, interoperability, and inclusive AI systems.

Additionally, it highlights the challenge of legacy systems in digitalisation and the need for investment in both technology and people.

Arguments

AI has been around for decades, but it's only started to become more mainstream and accessible in the last few years due to factors like affordable data storage, improved processing and enhanced connectivity.

Supporting facts:

• AI used to be restricted to university labs and big companies, but now it's accessible to everyday people, including students.

• The ability to store large quantities of data has improved due to big data centers and cheaper storage options.

• Processing of data has seen substantial improvements, but it is still a potential bottleneck.

• Improved connectivity allows people to access these data centers and run AI applications from almost anywhere.

Topics: AI, Data Storage, Processing, Connectivity

---

Limited connectivity is a major issue in developing markets, especially in the global south.

Supporting facts:

• In countries in Africa where there is limited access to 5G and fiber networks, the use of AI is expected to be restricted for several years.

• More investment is needed to improve connectivity infrastructure, including laying more fiber cables under the seas and building more data centers.

• Companies leading the generative AI revolution need to work closely with telecom operators to ensure that the necessary connectivity infrastructure is in place.

Topics: Connectivity, Developing Markets, Global South

---

Resilience in infrastructure, such as telecom infrastructure, is increasingly important and has proven to be vital during the COVID-19 pandemic

Supporting facts:

• The world relied heavily on digital connectivity to work and learn remotely during the pandemic, highlighting the importance of a resilient infrastructure.

• This resilience is easier to accomplish in countries with access to power and the means to invest in it.

Topics: COVID-19, Telecom Infrastructure, Resilience, Digital Connectivity

---

It's challenging to build resilience in infrastructure, particularly in the global south

Supporting facts:

• In big cities, it's easier because of the availability of multiple fiber cables.

• Connecting the next billion people is difficult not only because of the technology itself, but also because of the need for available power.

Topics: Global South, Resilience, Infrastructure

---

Telecommunication industry can gradually change out of legacy systems due to its nature of technology cycles

Supporting facts:

• In many countries, communication companies have switched off copper for fiber, introduced 5G, and modernized the core network

• A switch site which previously required 500 square meters of servers now only needs 20 square meters due to increased condensation and cloud capability

Topics: Telecommunication, Technology

---

AI plays a integral role in enhancing the reliability of the telecom networks

Supporting facts:

• Self-healing networks, which use AI, can reconfigure themselves in case of errors, thus reducing network failures and maintaining continuous operations

Topics: AI in Telecommunication, Network Reliability

---

AI is not just replacing people, it will replace some of repetitive jobs but it also augments the capability of other people

Supporting facts:

• In the last two years, they are hiring more people that have AI exposure

Topics: Artificial Intelligence, Job replacement, Job enrichment

---

Report

AI has become more accessible to everyday people, including students, in recent years. This shift has been driven by factors such as affordable data storage, improved processing capabilities, and enhanced connectivity. In the past, AI was mainly restricted to university labs and big companies, but now it is more readily available to a wider audience.

The affordability of data storage and the presence of big data centres have played a significant role in facilitating the accessibility of AI. The ability to store large quantities of data has improved, thanks to advancements in technology and the availability of cheaper storage options.

This has enabled individuals and organisations to accumulate and analyse vast amounts of data, which is essential for training AI systems. Furthermore, the processing of data has seen substantial improvements, although it still poses a potential bottleneck. The advancements in processing power have enabled faster and more efficient computation, making AI applications more viable and accessible.

However, further developments are needed in this area to ensure that processing capabilities continue to keep pace with the requirements of AI. Enhanced connectivity has also played a crucial role in increasing the accessibility of AI. Improved connectivity allows people to access data centres and run AI applications from almost anywhere.

This has significant implications for remote work and learning, as it enables individuals to connect to AI-powered systems and services regardless of their physical location. However, while the accessibility of AI has improved in many parts of the world, limited connectivity remains a major concern in developing markets, particularly in the global south.

In countries in Africa, for example, where access to 5G and fibre networks is limited, the use of AI is expected to be restricted for several years. To address this issue, more investment is needed in improving connectivity infrastructure, including laying more fibre cables under the seas and building additional data centres.

It is crucial for companies leading the generative AI revolution to work closely with telecom operators to ensure that the necessary connectivity infrastructure is in place. The COVID-19 pandemic has further highlighted the importance of resilient infrastructure, including telecom infrastructure.

The world heavily relied on digital connectivity to work and learn remotely during the pandemic, emphasising the need for a robust and reliable infrastructure. Resilience in infrastructure, particularly in the face of unexpected disruptions, is increasingly crucial. However, achieving this resilience is easier in countries with access to power and the means to invest in infrastructure.

In terms of bridging the digital divide, it is vital to ensure that connectivity and AI technologies are accessible to all, particularly in the global south. Closing the digital divide between developed countries and the global south is crucial for reducing inequalities and promoting sustainable development.

This requires not only affordable devices but also cost-effective ways to connect them. Efforts should be made to provide affordable technology solutions and improve connectivity access for underserved populations. The telecommunications industry is gradually transitioning away from legacy systems due to technology cycles.

Communication companies are switching from copper to fibre, introducing 5G, and modernising their core networks. These advancements are enabling more efficient communication and reducing the physical footprint required for communication infrastructure. AI also plays a crucial role in enhancing the reliability of telecommunications networks.

Self-healing networks, which utilise AI algorithms, can reconfigure themselves in case of errors, reducing network failures and maintaining continuous operations. This highlights the significant contributions AI can make to improve the reliability and resilience of critical infrastructure. Furthermore, AI is not just replacing jobs but also augmenting the capabilities of people.

While some repetitive jobs may be automated, AI systems also require skilled professionals to develop, deploy, and maintain them. Consequently, AI has led to an increased demand for individuals with AI expertise. This demonstrates the importance of investing in both technology and people to ensure successful AI implementation.

In conclusion, AI has become more accessible to everyday people, including students, due to factors such as affordable data storage, improved processing capabilities, and enhanced connectivity. However, limited connectivity remains a significant concern in many developing markets, particularly in the global south.

Building resilience in infrastructure, especially telecommunications infrastructure, is crucial, as highlighted by the COVID-19 pandemic. Investments in renewable energy and technology advancements, including 5G, offer promising opportunities for enhancing resilience and connecting more people. Closing the digital divide between developed countries and the global south is a critical goal that requires affordable devices and cost-effective connectivity solutions.

The telecommunications industry is gradually transitioning from legacy systems to improve efficiency, and AI plays a vital role in enhancing the reliability of telecommunications networks. Additionally, AI is not just replacing jobs but also augmenting the capabilities of individuals. Investment in both technology and people is necessary for success in the AI era.

Report

In the morning session at the Davos conference, the focus was on artificial intelligence (AI) and the importance of aligning infrastructure development with the rapid advancements in the AI sector. Experts from various fields came together for a panel discussion on how to promote global inclusion and competition in the field of AI.

The panel addressed the lack of attention given to the underlying elements of AI, such as the data, tools, and computing required for its development. They aimed to explore how infrastructure development can be aligned with the fast-paced progress in AI to foster inclusivity and competition on a global scale.

The panel consisted of Joris Poort, Lukas Biewald, Hatem Dovidar, and Emilia Stomenova-Du, each contributing their unique perspectives. Joris Poort, CEO of Rescale, stressed the importance of safe and fair engineering and research practices in AI development. Lukas Biewald, CEO of Weights and Biases, expressed concerns about the availability of essential components like chips, data centers, and 5G networks for the future of AI.

Hatem Dovidar, CEO of EAND and a member of the UN's Internet Governance Forum, highlighted the role of governments in promoting inclusivity and competition in AI. He emphasized the need for investments in resources, talent, and attention to create an environment conducive to AI development.

Emilia Stomenova-Du, the Minister of Digital Transformation for Slovenia, shared insights on the government's role in fostering inclusivity and competition in AI. The discussion then turned to the topic of interoperability as a crucial factor in achieving global inclusion. Emilia Stomenova-Du emphasized its importance and the challenges that come with it.

The seamless integration of data was also identified as vital for promoting inclusivity, with Lukas Biewald providing insights on its current status and potential future advancements. Another significant aspect of the session was the discussion on data access and its implications.

The Director-General of the World Trade Organization shed light on the advantage held by those in control of the data used to train AI models. This raised important questions about the sources of data, access, and control. Hatem Dovidar drew parallels between discussions on AI and the status of supply chains.

He emphasized the need for inclusive, sustainable, trusted, and resilient supply chains in the context of AI infrastructure. The session concluded with audience questions that further explored the topics discussed. The panelists addressed the topic of investment in machines versus investment in people, prompting reflections on society's priorities.

Emilia Stomenova-Du considered the potential synergy between investing in machines and investing in people, while Lukas Biewald shared insights from an investor's perspective, highlighting significant investments being made in machinery. Overall, the session highlighted the importance of sustainable, resilient, trusted, and inclusive infrastructure to support the development of AI.

It emphasized key considerations such as engineering and research practices, supply chain resiliency, interoperability, and seamless data integration. The panelists shared varying viewpoints on the balance between investing in machines and people, ultimately emphasizing the need to align AI advancements with societal needs and priorities.

Arguments

The need for sustainable and equitable practices in harnessing AI

Supporting facts:

• Data being generated at an exponential rate

• Growth in computational resources required to train AI models

• Linear growth in energy

• Potential and effective use of renewable energy sources

• Prospects for specialized and efficient computing capabilities

Topics: artificial intelligence, sustainability, equity, data infrastructure, energy, computing, public-private partnerships

---

There are significant innovation potential and investment incentives in developing AI and accessibility to data should be democratized.

Supporting facts:

• Advancements in AI like neural nets can revolutionize engineering and science

• Large scale data sets can produce significantly faster results in science and engineering.

• Companies are investing in making technologies more accessible.

• The cost of technologies decline over time, ensuring more accessibility.

Topics: AI, Data Inequality, Innovation, Investments

---

Cloud deployment allows refreshing of legacy infrastructure to hyperscalers or other types of cloud providers

Supporting facts:

• Cloud deployment ultimately allows you to push that sort of the challenge of refreshing legacy infrastructure to hyperscalers

Topics: Cloud Deployment, Legacy Infrastructure, Hyperscalers

---

The new challenges lie further up in the software layers in terms of how the technology is used

Topics: Software layers, Technology usage

---

Technology is seen as neutral and should be used to optimise human creativity and productivity.

Supporting facts:

• Around 25% of roles in the U.S. get re-skilled each year.

• AI techniques are helping engineers and scientists work faster and bring technologies to market faster.

Topics: Technology, Productivity, Creativity

---

Another powerful argument is the need for re-skilling.

Supporting facts:

• The working environment is dynamic with about a quarter of roles getting re-skilled each year in the US.

Topics: Re-skilling, Job Market

---

AI techniques are expediting drug development and enabling scientists to focus on pivotal tasks.

Supporting facts:

• There are drugs on the market today that have been developed with AI techniques.

Topics: Artificial Intelligence, Drug Development

---

The importance of innovation cannot be overstated.

Topics: Innovation

---

The man-machine symbiosis is very powerful and significant.

Supporting facts:

• Re-skilling allows workers to 'move up the stack' in value

Topics: Artificial Intelligence, Man-Machine Symbiosis

---

Report

The analysis explores various perspectives on harnessing artificial intelligence (AI) and its implications for sustainability, equity, innovation, and infrastructure. One key point emphasizes the importance of adopting sustainable and equitable practices in AI development. This includes the potential for using renewable energy sources and specialized computing capabilities to minimize the environmental impact and enhance efficiency.

Collaboration between the private, public, and academic sectors is highlighted as crucial for successful AI implementation. The rapid advances made by the private sector underscore the need for cooperation to ensure the responsible and ethical development of AI technologies. Additionally, universities play a vital role in driving technological development, and the involvement of the public sector is emphasized for effective control and governance.

The analysis also identifies the significant innovation potential and investment incentives associated with AI development. Access to data is recognized as a key factor in democratizing AI and ensuring its accessibility to a wider audience. Companies are investing in technologies that make AI more accessible, and as the cost of technologies declines over time, greater accessibility is expected.

The private sector's role in producing innovations and democratizing technology is emphasized. The private industry helps incentivize the democratization of technology through the downward cost curves of technologies, making them more affordable and accessible. Furthermore, the use of AI products in the private sector can lead to increased profits and economic growth.

Cloud deployment is highlighted as a means to refresh legacy infrastructure, particularly through collaboration with hyperscalers or other cloud providers. This approach is widely adopted by both private industry and government organizations, as it offers an efficient and easy way to stay updated on infrastructure requirements.

The challenges of technology lie in the software layers and how it is used to optimize human creativity and productivity. AI techniques are observed to help engineers and scientists work faster, resulting in faster product development and improved market conditions.

Additionally, the dynamic nature of the working environment necessitates reskilling, with approximately a quarter of roles in the US undergoing reskilling each year. AI techniques are acknowledged for their role in expediting drug development, enabling scientists to focus on critical tasks.

The importance of innovation in driving technological advancements is highlighted. The potential of man-machine symbiosis is recognized, especially in the context of reskilling, which allows workers to move up the value stack. Overall, the analysis conveys an optimistic view of the future of AI and its potential benefits for society.

The potential of AI in boosting efficiency, productivity, and improving health conditions is seen as a positive development. The need for collaboration, sustainable practices, and innovation is crucial in realizing these benefits while ensuring equity and responsible use of AI technologies.

Arguments

The development of Gen AI models is constrained by the availability of chips

Supporting facts:

• Over 90% of the compute on our platform happens on NVIDIA chips

• NVIDIA is the main producer of chips needed for training these models

Topics: Gen AI, Chip production, NVIDIA

---

He believes the energy bottleneck in model production is greatly exaggerated

Supporting facts:

• Data centres only account for a few percent of energy use in the United States

• The amount of GPUs is not constrained by energy, but by the ability to build these chips

Topics: Energy consumption, Model production

---

Data integration isn't typically seamless

Supporting facts:

• There is often overlapping pieces of infrastructure that need to work together, which is complex

• Models built today are different than those built years ago, meaning different infrastructures may be needed

• Changing hardware can require rewriting large parts of a platform

Topics: Data integration, Machine learning, Infrastructure

---

AI systems are becoming good at writing code

Supporting facts:

• AI systems can help with productivity

• Biewald found an AI system called Copilot helpful when he returned from leave

Topics: AI systems, Coding, Productivity

---

AI could potentially ease the process of data integration

Supporting facts:

• AI systems are good at certain tasks, one of which could be data integration

• Data integration has historically been an expensive and painful part of projects

• There is hope that AI could simplify the process of data integration

Topics: AI systems, Data integration

---

Large inequality in data access exists even among big tech companies

Supporting facts:

• Every large tech company is jealous of other tech companies' data sets

Topics: Data Access, Big tech companies

---

Data and types of data collected have inherent biases

Topics: Data collection, Bias

---

Dangerous and sensitive information can emerge once models are trained on all human knowledge

Supporting facts:

• Model could suggest ways to build weapons

• The model was good at suggesting how to build bioweapons

Topics: Artificial intelligence, Data security, Privacy

---

Companies need to make models and the data they're trained on publicly available

Topics: Data transparency, Open data

---

Regulations should insist companies reveal the data sets models were trained on

Topics: Regulations, Data transparency

---

AI has gotten incredibly good at extracting information from images, even poor quality images of documents

Supporting facts:

• AI applications are typically used to interact with legacy systems, which includes extracting information from documents

Topics: Artificial Intelligence, Legacy systems, Data extraction

---

Lukas Biewald supports investment in technology.

Supporting facts:

• Biewald indicates his approval of investment in technology

Topics: Investment, Technology, AI

---

Biewald sees a risk in AI applications

Supporting facts:

• He acknowledges there's clearly tons of risk herewith what AI can do.

Topics: AI, Risk Management

---

Report

The development of Gen AI models is facing a significant constraint due to the limited availability of chips. With over 90% of compute on the platform relying on NVIDIA chips, the shortage of these chips is hindering the progress of Gen AI.

NVIDIA is the main producer of chips that are essential for training these models. The scarcity of chips is negatively impacting the development and advancement of Gen AI. Regarding energy consumption in model production, Lukas Biewald believes that the energy bottleneck is exaggerated.

Data centres, which are responsible for model production, only account for a small percentage of overall energy use in the United States. Biewald argues that the constraint in model production lies not in energy availability but in the ability to build the necessary chips for producing these models.

Biewald sees potential for more diverse deployment of models, particularly in the inference side of model development. He highlights that companies like Qualcomm are creating interesting takes on the deployment of models. This suggests that there is room for innovation and diverse applications in the deployment of AI models.

Data integration proves to be a complex task as there are often overlapping pieces of infrastructure that need to work together. Building models today requires different infrastructures compared to models built years ago, which further complicates the data integration process.

Additionally, switching to different hardware can necessitate rewriting a substantial portion of a platform. This indicates the complexities involved in ensuring seamless data integration. AI systems have shown promise in enhancing productivity. Biewald found an AI system called Copilot to be helpful when he returned from leave, suggesting that AI systems are becoming adept at writing code.

Furthermore, AI systems could potentially simplify the process of data integration, which historically has been expensive and painful for projects. The hope is that AI can alleviate the challenges associated with data integration. Large inequality in data access exists, even among big tech companies.

Biewald asserts that every large tech company is jealous of the data sets owned by other tech companies, highlighting the stark disparities in data access within the industry. Data collection and the types of data collected have inherent biases. This bias can pose challenges when training AI models and can affect the accuracy and fairness of the models' outputs.

Being aware of and mitigating these biases is crucial for developing ethical and unbiased AI systems. Trained AI models have the potential to unearth dangerous and sensitive information. Biewald cites examples of models suggesting ways to build weapons and bioweapons based on human knowledge.

This highlights the need for careful consideration of the information and knowledge used to train AI models and the potential risks associated with unrestricted access to such models. Biewald emphasises the importance of data transparency and open data. He argues that companies should make their models and the data they are trained on publicly available.

This promotes accountability, facilitates collaboration, and allows for continuous improvement and validation of AI models. Regulations should play a role in ensuring data transparency and open access. Biewald suggests that regulations should require companies to reveal the data sets their models were trained on.

This would provide insights into the sources of information used to make important decisions and help maintain trust and transparency within the AI industry. AI applications have proven to be particularly effective in extracting information from legacy systems, such as extracting information from documents.

This capability can greatly facilitate interoperability with legacy systems, improving efficiency and productivity in various sectors, including healthcare. Biewald supports investment in technology, highlighting the importance of embracing innovation and fostering advancements in the field of AI. However, he also acknowledges the risks associated with AI applications.

He recognises that there are significant risks involved in what AI can do, and these risks should be carefully managed and mitigated. Biewald disagrees with a generalized approach to AI regulation. He criticizes the idea of applying the same regulations used for industries like aviation to AI, which encompasses multiple applications and functions.

He argues that regulations should be tailored to individual AI applications, taking into account their specific characteristics and potential risks. In conclusion, the development of Gen AI models is hampered by the limited availability of chips, while the energy bottleneck in model production is deemed exaggerated.

There is potential for more diverse deployment of models, and AI systems are proving beneficial in improving productivity and simplifying data integration processes. However, challenges remain in data integration, data access inequality, biases in data collection, and the risks associated with unrestricted access to trained models.

Biewald emphasises the importance of data transparency, open data, and regulations based on individual applications. He also supports investment in technology while recognizing the need for careful management of the risks associated with AI applications.