Currently, the AI world is changing at a breath-taking speed, making it difficult to comprehend the basic dynamics. In fact, the disruptive power of so-called foundation models (sometimes also base models) is already visible today and can hardly be ignored, especially by decision makers in industry. The first phase of the digital transformation has already found their victims in the AI transformation. For example, 52% of the Fortune 500 companies have disappeared since the year 2000 and it is not expected that the adaptation difficulties will stop, as the development speed of AI-based applications has increased dramatically.
Andrew Ng summarized the pressure already in 2020 to adapt as follows: “In the past, a lot of S&P 500 CEOs wished they had started thinking sooner than they did about their Internet strategy. I think five years from now there will be a number of S&P 500 CEOs that will wish they’d started thinking earlier about their AI strategy.” In fact, the authors believe that there is not that much time left. This is true: The diffusion speed of so-called generative AI is so great and its immediate, beneficial applications so broad that we will see concrete, competitive examples of implementation at the industrial level in the next few months.
The following two current examples may serve to illustrate the dramatic consequences that are to be expected:
- on 2.5 of this year, the CEO of IBM Arvind Krishna announced that 7800 open positions at the company will not be filled, because it is assumed that the productivity of these positions can also be achieved through AI. https://arstechnica.com/information-technology/2023/05/ibm-pauses-hiring-around-7800-roles-that-could-be-replaced-by-ai/
- on the same day, the stock price of the American provider of learning support plummeted by over 40% when the company had to admit its revenue declines as a pole of the use of ChatGPT. https://www.cnbc.com/2023/05/02/chegg-drops-more-than-40percent-after-saying-chatgpt-is-killing-its-business.html
The speed of developments can be explained by the fact that current AI applications are encountering an economy that is already largely digitalized. ChatGPT was released to the general public in November 2022 and had made it to 100 million registered users within eight weeks. Microsoft announced on March 17 that all users of the MS 365 suite will have access to the AI-based “Co-pilot,” immediately reaching 340 million users.
This speed of digital dissemination is unparalleled in the history of technology. The telephone was introduced in 1878 and took 75 years to reach 100 million users. The mobile phone, launched in 1979, took another 16 years, and the World Wide Web achieved this number of users in eight years. The social network Facebook, launched in 2004, needed only 4.4 years to reach the magical 100 million threshold, Instagram launched in 2012 made it in 2.2 years. But the diffusion speed achieved today is indeed breath-taking, and this also makes the quotes mentioned at the beginning understandable.
Dr. Andreas Liebl, managing director of the appliedAI Initiative gave a presentation on the industrial application of generative AI at the CTO Forum on Tuesday, May 2. appliedAI is Europe’s largest initiative for the application of leading edge trustworthy AI technology with the vision to advance Europe’s industry to compete in the age of AI. appliedAI was formed as trustworthy initiative that acts both as enabler and innovator, shaping a future that we desire to live in.
He explained the current impact of Large Language Models on what applications are economically viable at the industrial level. The currently available foundation models have indeed fundamentally changed the economic assessment for the operational use of AI, also because the quality of the output has improved drastically in a very short time.
As the Samsung case, which became public on April 6, shows, it is necessary for companies to deal with their own intellectual property in connection with the use of public LLMs in particular. Employees of the Korean tech giant published trade secrets when source code was to be corrected by ChatGPT.
In the application of AI-based systems, a significantly greater degree of understanding of one’s own trade secrets, one’s own know-how is required than was previously the case around. In many companies, the topic of know-how protection has been rather neglected up to now. In a competitive environment based on knowledge, this can quickly have significant consequences, as shown by the know-how theft cases that are soon to be published on a weekly basis worldwide.