You make a smoothie for your friends. Already mixed with various fruits and yogurt, your friend Ruchir comes with a ripe apple and gives it to you so that you can finish your refreshing masterpiece. After the drink is finished, you can almost still smell the apple note. Before taking the first sip, Ruchir says, “I changed my mind, I have to go and I want the apple back.” They say, “Oh,I’m sorry, but that’s just impossible.“We’ll come back to this story in a moment and explain how it relates to ChatGPT and trustworthy AI.
As the world of artificial intelligence (AI) continues to evolve, new tools like OpenAI’s ChatGPT have gained attention for their conversational features. However, I also understand the importance of assessing the inherent risks before directly implementing them in our organizations. In this discussion, I examine the risks and challenges of ChatGPT in an enterprise context, which requires a careful approach to its implementation. Additionally, I will highlight the importance of implementing IBM Watsonx to ensure reliable AI solutions.And when in doubt, I advise you to use the same common sense you have always used when using new online services.
The development of artificial intelligence tools
ChatGPT leverages the enormous power of GPT-3 and GPT-4, a new class of large, “gigantic” and very popular language models used in various artificial intelligence applications. ChatGPT allows users to ask questions, generate texts, create email versions, discuss code in different programming languages, translate natural language into code, and much more. It features a high-quality conversational chatbot that aims to provide consistent and contextual answers.
ChatGPT is the ideal tool to explore creative writing, generate ideas and interact with artificial intelligence.It’s free for everyone and an advanced version is available for ChatGPT Plus subscribers. A chatbot’s ability to remember previous conversations increases its interactivity and engaging experience.
Although ChatGPT has gained attention and popularity, it faces competition from other AI chatbots and natural language processing (NLP) systems. For example, Google developed Bard, its AI chatbot that runs on its own voice engine called PaLM 2. Likewise, Meta recently released its impressive LLaMA2 model.As the AI-based chatbot space continues to evolve, there will definitely be increased competition and the emergence of new players. To find the best solutions for your business needs, you need to stay up to date with developments in this industry.
Before companies look for alternative solutions, it is important to be aware of the potential risks and challenges of using ChatGPT directly. Remember that the history of the Internet has shown the emergence and development of new services (e.g. Google search, social media platforms, etc.), highlighting the importance of privacy and data ownership in the enterprise.With that in mind, here are the most important factors toconsider:
Security and data leakage
When sensitive third-party or company-internal information is entered into ChatGPT, it becomes part of the chatbot’s data model and can be shared with others who ask relevant questions. This can lead to data leaks and violations of the organization’s security policies.
Example
: Plans for a new product your team is launching for a customer, including confidential specifications and marketing strategies, should not be shared with ChatGPT to avoid the risk of data leaks and potential security vulnerabilities.
Confidentiality and Privacy
Similar to the previous point, sharing confidential information aboutcustomers or partners may violate contractual agreements and legal requirements to protect that information. If ChatGPT’s security is compromised, confidential content could be leaked, which could negatively impact an organization’s reputation and lead to criminal prosecution.
Example: Suppose a healthcare organization uses ChatGPT to answer patient questions. If sensitive patient information, such as medical records or personal health information, is disclosed to ChatGPT, it could potentially violate legal obligations and patient privacy rights protected by laws such as HIPAA (Health Insurance Portability and Accountability Act) in the United States.
Intellectual property issues
Owning code or text generated by ChatGPT can be complex. The terms of use state that the output belongs to the input provider, but problems may arise if the output contains proprietary data from other inputs. Copyright issues can also arise when ChatGPT is used to generate written material based on copyrighted property.
Example: Generating written material for marketing purposes that results in copyrighted content from external sources without proper attribution or permission may potentially infringe the intellectual property rights of the creators of the original content. This can have legal consequences and a loss of reputation for the company.
Compatibility with open source licenses
If ChatGPT uses open source libraries and incorporates this code into its products, it could potentially violate open source software licenses (OSS) (e.g. GPL), which would lead to legal complications for organizations.
Example: If a company uses ChatGPT to generate code for software and the source of the training data used to train GPT is unclear, there is a risk of possible violation of the open source license terms associated with that code are connected. This can lead to legal complications, including license infringement claims and possible lawsuits from the open source community.
Limits to the development of artificial intelligence The terms of use for
ChatGPT state that it may not be used to develop other artificialintelligence systems. Using ChatGPT in this way could hinder future AI development plans if a company operates in this area.
Example: A company specializing in speech recognition technology is exploring the possibility of improving its existing system by integrating ChatGPT’s natural language processing capabilities. However, ChatGPT’s terms of service clearly state that it cannot be used to develop other AI systems.