Generative AI: Opportunities Without Risks?
The topic of generative AI (GenAI) is currently on everyone’s lips. New possible applications are presented and discussed on a daily basis and almost every software provider – regardless of the industry – is advertising corresponding “intelligent” features.
The widespread use of generative AI is first and foremost due to its remarkable ability to increase efficiency and creativity in various areas. Both individuals and companies benefit from GenAI, as it is able to automate repetitive tasks, generate high-quality content (text, audio, image, video, etc.) and create insightful data analyses, for example. In addition, the easy accessibility and user-friendliness of GenAI platforms contributes to their popularity and spread, as it enables people without technical expertise to use GenAI easily.
The rapid spread of generative AI (GenAI) has fundamentally changed numerous industries and everyday life by enabling the creation of sophisticated content and the automation of complex processes. Professionals in creative fields such as writing, design and music are using GenAI tools to brainstorm ideas, design content and create artwork, significantly speeding up their workflows and expanding their creative horizons (see Adobe Firefly – A new dimension of generative AI). In marketing, GenAI tools can generate tailored advertising and personalized customer interactions, significantly increasing engagement and conversion rates. In customer service, chatbots with GenAI can handle requests with remarkable accuracy and empathy, increasing customer satisfaction and reducing operating costs (see Marketing power of generative artificial intelligence).
In addition, the ongoing integration of GenAI into everyday applications has made it less and less dispensable for both personal and professional use. The combination of practicality, accessibility and continuous improvement makes GenAI a ubiquitous tool in modern society and ensures broad acceptance and integration into various aspects of daily life.
Clearly, GenAI offers great opportunities and possibilities, but it also undoubtedly carries some risks that should not be underestimated and also raises ethical and societal issues, such as the potential for job displacement, privacy and copyright issues or the spread of misinformation through deepfakes.
Potential Risks Associated with the Use of GenAI in Enterprises
As already indicated, the use of GenAI also harbors considerable risks and dangers on closer inspection. These problems primarily concern the use of GenAI in companies, where the careless use of GenAI solutions can have devastating consequences. In the following, we will therefore take a closer look at some of the potential threats to companies.
1. Copyright Infringements
The use of generative AI in companies harbors considerable copyright risks. AI models rely on extensive data sets for training, which may contain copyrighted material. If the training data contains copyrighted content without proper authorization, the resulting AI output could infringe intellectual property rights and expose the company using the AI output to legal liabilities. For example, several record companies (Sony, Universal and Warner) have recently sued the AI song generators Suno and Udio for copyright infringement.
In addition, generative AI can inadvertently produce results, such as images, that are very similar to existing copyrighted works, which can lead to potential copyright infringement lawsuits. To make matters worse for companies in the US, the US Copyright Office recently ruled again that artwork created by artificial intelligence is not protected by copyright, confirming its practice of not registering AI-generated art as a copyrighted work. These points are particularly concerning in areas of content creation, and in the advertising and media industries, where originality and licensing rights are crucial.
Risk mitigation strategies
- The central question regarding possible copyright infringements is which training data was used to train the GenAI model used. Can it be ensured that the training data did not contain any copyrighted material (text, images, audio, etc.)? To the best of our current knowledge, Adobe with its Firefly product is the only provider that is considered commercially safe in the field of AI image generation, as the corresponding models were trained exclusively with Adobe Stock images and freely licensed works. Adobe even indemnifies its corporate customers against liability for IP infringements. In addition, Adobe makes it possible to train custom models with your own company data.
- Introduction of strict data governance guidelines for training custom models.
- Establishment of defined review and approval processes at legal and content level for AI-generated assets.
In this vide0, Ben Derico from “BBC Click” illustrates the threat that generative AI can pose to artists.
2. Loss of Reputation
The use of generative AI in companies carries a significant risk of reputational damage. If an AI system produces biased, offensive or inaccurate content, it can damage a company’s reputation as well as its brand and undermine public trust in the company.
For example, a generative AI model used in customer service, such as a chatbot, could produce inappropriate or insensitive responses, leading to customer rejection and negative publicity. One example of this is the AI chatbot used by parcel delivery company DPD, which started swearing in a chat with a customer and criticizing the company itself.
Similarly, if an AI system spreads misinformation or creates deepfakes, the company could be seen as irresponsible or unethical. In addition, any perceived lack of transparency or accountability in AI decision-making processes can further damage a company’s credibility
Risk mitigation strategies
- Establish defined review and approval processes on a legal, content and ethical level.
- Ensuring strict testing procedures and continuous monitoring.
Parcel delivery firm DPD have replaced their customer service chat with an AI robot thing. It’s utterly useless at answering any queries, and when asked, it happily produced a poem about how terrible they are as a company. It also swore at me. 😂 pic.twitter.com/vjWlrIP3wn
— Ashley Beauchamp (@ashbeauchamp) January 18, 2024
Tweet from Ashley Beauchamp about the unexpected behavior of the DPD chatbot.s.
3. Loss of Know-how
The integration of generative AI in companies carries the risk of losing important know-how and expertise in the long term. As AI systems automate tasks and processes, there is a risk that human employees will rely too much on these technologies, leading to a gradual erosion of essential skills and institutional knowledge. This dependency can be particularly detrimental in complex problem-solving scenarios where human intuition and experience are crucial. In addition, AI models generally hide their underlying mechanisms and decision-making processes (black box), making it difficult for employees to understand how AI works and learn from it.
Risk mitigation strategies
- Continuous investment in education and training programs.
- Ensure a balanced approach to AI and human collaboration.
- Keep and maintain comprehensive documentation on all AI systems used within the company and their functions.
Summary and Conclusion
Generative AI (GenAI) has quickly developed into a key technology that is increasingly transforming entire industries by increasing efficiency and automatically generating high-quality content (text, images, videos, etc.). But everyday life is also being increasingly influenced by the growing private use of GenAI applications.
The potential applications of generative AI are almost limitless and undoubtedly hold enormous savings and revenue potential for companies. Despite all the prevailing euphoria, however, it is essential to consider the risks associated with the use of GenAI applications for companies. These risks essentially relate to three areas:
1) The risk of copyright infringement
Generative AI results (text, images, audio, etc.) may infringe intellectual property rights if the training data is based on copyrighted content. In addition, generative AI can inadvertently produce results that are very similar to works already protected by copyright. Both points could lead to lawsuits for copyright infringement.
2) The risk of loss of reputation
If an AI system produces biased, offensive or inaccurate content, this can significantly damage a company’s reputation and undermine public trust in the company. Such damage can then only be remedied with immense effort, if at all.
3) Risk of loss of know-how
The continuous use of AI systems to automate processes leads to dependencies and the loss of specialist knowledge among the employees concerned. A failure of the systems can then lead to a long-term standstill of the corresponding processes.
To mitigate the GenAI risks described above, companies should follow a few basic procedures:
- Every planned use of GenAI systems should be scrutinized very critically to determine whether it actually makes sense. It is also important to clarify the potential risks associated with the use of GenAI in the given use case and what specific measures can be taken to minimize the risk. In addition to a cost-benefit analysis, an application-specific opportunity/risk analysis should therefore always be carried out. The uncontrolled use of GenAI solutions in the company should be emphatically prevented.
- As a rule, the use of GenAI requires mandatory defined review and approval processes, both in terms of content and on a legal and ethical level. This applies in particular if GenAI is used to generate content for external corporate communication.
- For use in the company, it would definitely be important to understand how a GenAI application works and to know the underlying training data. While copyright issues are at the forefront when creating content such as images or texts, when using GenAI in decision-making processes, it should be possible to understand the basis on which the system generates the corresponding results. The fact that many providers sell their GenAI-based solutions as a black box does not inspire much confidence from a business perspective. Companies should therefore check in advance what information the GenAI provider makes available regarding functionality and training data. An assessment must also be carried out with regard to the threat of copyright infringements. Reputable providers who describe themselves as commercially secure ideally provide an indemnity against liability for IP infringements.
The use of generative AI offers impressive possibilities and opportunities – both in the private and business environment. However, there are also some significant problems and risks, which are often pushed into the background in the current euphoria. In addition to the criminal use of generative AI to create almost perfect emails for phishing attacks or the targeted creation and dissemination of misinformation through deepfakes, the use of GenAI also raises ethical and social issues. These include, for example, the possible displacement of jobs through automation or discrimination based on biased training data. Especially when using GenAI systems in companies, those responsible should think about potential risks in advance and prepare themselves accordingly. Otherwise, there is a risk of lawsuits for copyright infringement, loss of reputation or a significant loss of employee know-how.
