New Ethical Issues Brought by Generative Artificial Intelligence

The rapid development of generative artificial intelligence urgently requires ethical safeguards for data, scale, responsibility, copyright, and misinformation risks. The speed of advanced AI development is lightning-fast, but the accompanying risks are equally swift.

The situation has become so complex that experts find it difficult to anticipate the risks.

Although most leaders are increasingly focusing on GenAI applications in the coming months, they also have doubts about the associated risks, such as data security issues and result biases.

"While this technology can lead to breakthroughs, accelerate scientific progress, and improve learning effectiveness, it is not without risks," said Mark Suzman, CEO of the Bill and Melinda Gates Foundation.

Let's start with the data

Consider this fact: a renowned generative AI model creator stated, "It collects personal information such as names, email addresses, and payment information when necessary for business purposes."

Recent cases have shown that without guidance frameworks, things can go wrong in various ways.

• Italy expressed concerns about the illegal collection of personal data, citing "no legal basis to prove the large-scale collection and storage of personal data for algorithm 'training,' which forms the basis of platform operations."
• The Personal Information Protection Commission of Japan also issued warnings, demanding minimal data collection for training machine learning models.
• HBR (Harvard Business Review) echoed concerns about data security issues and result biases.

Since generative AI models are trained on almost all internet data, we can say they are a small part hidden within these neural network layers. This emphasizes the need to comply with data privacy regulations and not use user data for model training without their consent.

Recently, a company faced privacy breaches and heavy penalties for scraping selfies from the internet to build facial recognition tools.

However, data security, privacy, and biases existed even before the era of generative AI. So, what has changed with the introduction of generative AI applications?

Well, considering the scale of model training and deployment, some existing risks will only become greater. Let's understand how to handle them.

Scale - A double-edged sword

Understanding the internal workings of these massive models becomes more important to trust their responses. In Microsoft's words, these emerging risks are because LLM "aims to generate text that appears coherent and contextually appropriate, rather than adhering to factual accuracy."

As a result, these models may produce misleading and incorrect responses, often referred to as illusions. They may appear when the models lack confidence in their predictions, leading to the generation of inaccurate or irrelevant information.

Furthermore, prompts are how we interact with language models; now, malicious actors can inject prompts to generate harmful content.

Accountability when AI goes wrong?

Using LLM raises ethical questions about accountability and responsibility for the outputs and biased outputs generated by these models, which is common across all AI models.

High-risk applications, such as in healthcare, exacerbate the risks - think about the impact of incorrect medical advice on patient health and life.

Most importantly, companies need to establish ethical, transparent, and responsible approaches to developing and using generative AI.

Copyright infringement

Since these large-scale models are built on global materials, they are highly likely to consume creations such as music, videos, or books.

If copyrighted data is used to train AI models without obtaining necessary permissions, credits, or compensating the original creators, it can lead to copyright infringement and potentially result in severe legal disputes for developers.

Deepfakes, misinformation, and manipulation

There is a phenomenon that can potentially create chaos on a large scale - deepfakes. Want to know what deepfake capabilities can lead us into?

They are synthetic creations - text, images, or videos - that can digitally manipulate facial appearances through deep generative methods.

The result? Bullying, misinformation, prank calls, revenge, or fraud - none of which align with the definition of a thriving world.

This article aims to make everyone aware that artificial intelligence is a double-edged sword - it's not just magic that works for important initiatives; bad actors are also part of it. That's why we need to be vigilant.

Security measures

Take the recent example of a fake video news that highlighted a political figure's withdrawal from an upcoming election.

What could be the possible motives? - you might wonder. Such misinformation spreads like wildfire and can significantly influence the direction of the electoral process.

So, what can we do to avoid becoming victims of such false information?

There are various defense lines, let's start with the basics:

• Approach everything you see around you with skepticism and doubt.
• Set your default mode to "this might not be true" rather than blindly believing everything. In short, question everything around you.
• Verify potentially suspicious digital content from multiple sources.

Is pausing development the solution?

Renowned AI researchers and industry experts such as Yoshua Bengio, Stuart Russell, Elon Musk, Steve Wozniak, and Yuval Noah Harari have also expressed their concerns and called for a pause in the development of such AI systems.

There is a common concern that the race to build advanced AI, especially AI that rivals generative AI, could quickly get out of control.

Making progress

Microsoft recently announced that it will protect buyers of its AI products from copyright infringement as long as they comply with guardrails and content filters. This is a significant reassurance and demonstrates the intention to take responsible action to address the consequences of using their products - one of the core principles of an ethical framework.

This will ensure that authors retain control over their rights and receive fair compensation for their creations.

This is a great step in the right direction! The key is to see to what extent it addresses the concerns of authors.

What's next?

So far, we have discussed the key ethical implications associated with the proper use of technology. However, the risks arising from the successful application of this technological advancement result in job losses.

There is a fear that AI will replace a significant portion of our jobs. McKinsey recently shared a report on what future work might look like.