You've found us in English! The English version of TechSavvy.media is currently only available in a beta version. This means, among other things, that the majority of articles are machine translated. We hope you'll still want to stick around a little longer

This post is also available in: English

GENERATIVE AI

This post is also available in: English

The rapid advancements in generative AI models (GenAI), particularly large language models (LLMs), are reshaping the dynamics of human-machine interactions [12]. This transformative power has become increasingly evident to the public through the popularity of services like OpenAI’s ChatGPT and Google’s Bard. Notably, ChatGPT has had an astounding growth, attracting 100 million users within a mere two months [4]. This all testaments to the profound impact GenAI has already had but also the untapped possibility for unlocking the human potential in the future.

Dive deeper into the trends with our “Global Tech Trends” podcasts series:

GenAI

GenAI refers to a branch of advanced artificial intelligence technologies that focusses on creating and generating new and original content by utilizing large neural networks trained to recognize patterns and relationships in large datasets [13].

While the concept of GenAI is not new, the distinguishing factor lies in the unprecedented scale at which these new models have been trained. OpenAI’s introduction of ChatGPT, built on a staggering 175 billion parameters, is effectively encompassing a substantial portion of textual data available on the internet prior to 2021 [5]. This solidifies the emergence of LLMs, as a prominent term in this realm.

Why is this interesting?

GenAI is poised to have profound implications for businesses and society at large. The ability of GenAI models to produce realistic and creative content is driving transformative changes across various industries already today [12]. From general business efficiency, content creation, to healthcare and customer service, but this is only the beginning [2][7].

In the future GenAI will evolve in numerous ways. Text-based models will go from helping with drafting, researching and planning, to having the ability to write deeper and more engaging writings, as it will understand factors like psychology and human creative processes [2]

Visual AI will go from creating images based on natural language prompts, to being able to create images and videos about anything, to the extent that it will be difficult to tell if it is real or not. 

In music and sound, it will have the ability to create real-time adaptive soundtracks, and be a valuable tool to songwriters as they explore new creative ways [2]. It will also improve computer-generated voice, giving GenAI expressiveness and emotion close to that of humans, new ways of real-time translations, voice overs and more. 

In design processes, it will be able to assist with prototyping, designing new and unique products, speeding up the design and testing process, and giving the potential to create optimized never seen before designs with new functionality or abilities [2]

Even in video games or the metaverse, GenAI will have an impact, as it can conceptualize and build endless and immersive environments, and lifelike non-players or avatars that can interact like a human freeing up time for developers to work on the stories [2]

In the healthcare and pharmaceutical sector, 30% of all new drugs and material will be discovered using GenAI (0% today). Saving both time and costs [7].

The professional possibilities will be endless, from executing small tedious tasks, to big efficiency gains, transforming entire industries [12], and even creating new ones all together. Over the coming years, we will continue to see GenAI develop at an accelerated pace, being capable of carrying out an ever-growing number of tasks: ultimately augmenting our human skills, adding to and developing them – unlocking our human potential and efficiency in ways we cannot even imagine [2]. It is important to note, however, that GenAI also holds the potential to heighten the risks of both cyber-warfare, disinformation and deepfakes [21] As GenAI grows, it will continue to disrupt more industries and even our everyday lives, creating frustrations, concerns and the feeling of being threatened. It is therefore up to all of us, to not perceive this as a threat, but as a helping tool we need for unlocking our future potential.

How far are we?

Policy: Given the potential impact and implications, regulation is necessary to address ethical concerns, ensure privacy and data protection, mitigate potential biases or discriminatory outcomes and maintain accountability [3]. The latter sentiment is mirrored not only by public actors, but also by the private sector, as the Institute for the Future of Life alongside industry experts published an open letter in early 2023 to call for a pause of giant AI experiments [6]. However, regulation in the US on GenAI is still quite nascent, congress has passed several bills that focus on governing AI but most measures either seek to explore the topic or only affect how the government itself are using it [18][19]. Senator Chuck Schumer is working on a framework to regulate AI on a national level, but in general lawmakers on a global scale are struggling to keep up with the developments of the technology [17]. In California lawmakers seek to be first in regulating business uses of GenAI and recently introduced A.B. 331, which is a framework for the regulation of “Automated Decision Tools”. It aims at monitoring how employers and industries use these tools. Compared to previous bills this is more far-reaching, as it targets discrimination from AI in everything from employment, financial services, education and more [19].

Entrepreneurship: Y Combinator’s Summer 2022 batch included 11 startups focused on GenAI. However, in just six months, their Winter 2023 (W23) cohort saw a significant increase, with 59 GenAI start-ups, comprising 22% of the total 272 companies in the cohort. Garry Tan, Y Combinator’s president and CEO, stated in February that generative AI is “the new frontier” [8][9]. Y Combinator is a pioneering Silicon Valley based accelerator that each year invests in a large number of start-ups, to go through their intensive three-month accelerator and coaching program, with the aim to get them ready for pitching to US investors. By now they have funded 4.000 start-ups with a combined valuation of 600 billion USD. A few of the start-ups that later had huge success are Airbnb, Stripe, Dropbox, Instacard, Reddit, Flexport, Cruise and many more [14][15][16].

Investment: In 2022, GenAI garnered substantial investor interest, with USD 2.65 billion invested across 110 deals. Most companies had not raised Series A funding or received outside equity. Generative interfaces experienced the most momentum, with a remarkable 692% funding increase from 2021 to 2022 [8][9]. In Q1 of 2023, generative AI start-ups have raised roughly USD 1.7 billion across 46 deals [10] marking the ongoing interest in the field. 

Corporate: The applications are endless and the potential massive, but for a corporate looking to design, deploy, and scale GenAI it is important to build the right governance framework and ensure responsible use [3]. There are two primary implementation approaches depending on the intended application. The first option is Model-as-a-Service via API integration, using externally managed larger models like ChatGPT, acting as a universal tool. The second is an open-source software solution to internally run smaller models, focused on very specific tasks [1].

Academia: In efforts to capitalize on the rapid AI innovation and supporting further research, the US government is opening seven new AI labs. Aimed at advancing the development of AI within six research agendas; 1) trustworthy AI, 2) intelligent agents for cybersecurity, 3) climate-smart agriculture and forestry, 4) neural and cognitive foundations of AI, 5) AI for decision-making, and 6) AI-augmented learning. The laboratories will be an interdisciplinary collaboration between top AI researchers supported by co-funding from different US governmental departments. They will be based out of seven different universities around the US; 1) University of Minnesota, 2) University of Columbia, 3) UC Santa Barbara, 4) Carnegie Mellon University, 5) University of Buffalo, 6) University of Illinois and 7) University of Maryland [11].

TechPlomacy perspectives

The rapid development and expansion of generative AI systems has quickly risen to the top of the international agenda, with policy makers and industry experts moving to address the societal opportunities and challenges of AI. For example, in May 2023 the G7 has called for developing and adopting technical standards to keep artificial intelligence (AI) trustworthy, in line with democratic values. Many countries – including geopolitical powers – compete to reap the potentially enormous economic and commercial benefits of AI. At the same time, the rapid development raises a number of challenges and risks, including to cybersecurity and democratic credibility [20]. One particular challenge raised by many is AI’s role in development and spread of disinformation. As information products can rapidly be created, deployed and exist in an ‘accountability vacuum’ on social media (e.g. via anonymous profiles and more), the technology holds a risk for nefarious use, and could potentially become a tool used for malicious actors aiming to destabilize society, sow division and undermine democratic dialogue. Experts [21] worry, that ultra-realistic deepfake-material can be expediently done, and target political opponents or democratic structures by undermining the credibility of news, politicians and government by flooding the news sphere with fake alternatives. On the other hand, some experts also believe that AI can be an effective in countering cyberthreats and disinformation campaigns strengthening our ability to monitor, analyze and respond to threats. As AI continues to become more prevalent, it is important to discuss what the appropriate safeguards are against this development, and how to develop and deploy AI in a responsible way. Governments and political influencers are already working towards cautionary legislative approaches. However, as legislation processes typically proceeds slower than the speed of GenAI development, several countries, organizations and businesses have proposed a voluntary code-of-conduct. [22]

Let’s connect

Please reach out to Cathja Windbæk Lind catlin@um.dk for any inquiries. We, at ICDK Silicon Valley, offer our services to both corporates, SMEs and academic partners looking to dive further into the area of generative artificial intelligence.

Sources

[21]: Professor Hany Farid, Prof. UC Berkeley, in conversation with Tech Embassy.

Dive deeper into the trends at the Global Tech Trends conference:

Global Tech Trends conference on 11.SEPTEMBER

Innovation Centre Denmark launched our Global Tech Trends platform in collaboration with TechSavvy.media, which gathers 12 unique descriptions of technological trends from all over the world. You can learn more from the people behind the reports at our Global Tech Trends conference on September 11th.

WHEN

11TH OF SEPTEMBER, 11:30 AM - 5.30 PM

WHERE

STRANDGADE 25D, 1401 KØBENHAVN