Generative AI

• There is now intense hype around Generative AI models
• This technology will disrupt many sectors of the economy…
• … but raises important social and ethical considerations

The past year has seen a surge in interest in artificial intelligence (AI) and so-called generative models. These are machine learning models which can produce new content including text, images and music – something which until recently was considered to be the unique purview of humans.

Major newspapers are writing opinion articles on the technology; social media feeds are being inundated with discussions. But why now?

From a technical perspective, there has been a smooth progression of what is possible over the past few years. What has really changed is the ability for a non-technical audience to use this technology. Anyone who knows how to use a website – from entrepreneurs, to content creators – can now access and interact with generative models. We are seeing a dramatic expansion in the use cases being actively explored.

With any nascent technology it is hard to predict the various ways it will ultimately end up being used productively. There are clear implications for the big tech sector, and for industries manufacturing the advanced chipsets on which generative AI relies. But many more sectors besides could also see a significant impact.

In media, generative AI opens up the potential to produce content quickly and at lower cost. Music generated from AI has already started to appear on streaming platforms. Generative AI could be a powerful tool for education if used in the right way, though much of the initial debate has focused on fears of rising plagiarism. In consumer and retail, the technology promises the ability to tailor messages more tightly to individual consumers. And in pharmaceuticals and healthcare, while the impact has been muted so far, there is potential for generative AI to support in areas such as drug discovery.

As with all technological developments, there is a possibility of generative AI techniques being misused. Key ethical issues worth considering include:

1. Bias: Generative AI models can perpetuate and even amplify existing biases in the data they are trained on. For example, a model trained on a biased dataset of news articles might generate text that reflects those biases. This could perpetuate harmful stereotypes and discrimination
2. Privacy: Generative AI models can be used to generate realistic synthetic data, raising concerns about the protection of individuals’ privacy
3. Misuse: Generative AI models can be used to create deepfakes or other manipulated content, which can be used to spread misinformation or cause harm
4. Impact on Employment: Generative AI models can be used to automate the creation of content, raising concerns about the impact on employment in industries such as media and entertainment
5. Lack of Transparency: Generative models can be complex and hard to interpret, making it difficult to understand how they arrived at a certain output, which can lead to lack of trust in the model and its decisions
6. Inclusion: Generative models might not be accessible to everyone. This could lead to issues of social and economic inequality, as individuals who are not able to access these tools might be at a disadvantage

Still not convinced of the power of generative AI? Well, please note that the above numbered bullet points were in fact written by ChatGPT, a recently released model from OpenAI. Large sections of the video accompanying this piece were also entirely produced by AI. Check it out to see what’s now possible at very low cost.

While this type of technology is not yet perfect, it is already an extremely useful tool for anyone creating content. For more on the disruptive potential of this technology, and a deeper dive into the ethical considerations (this time written by a human), take a look at our report.

Would you like to know more? Click here* to read a free-to-view version of the report.

To find out more about HSBC Global Research or to become a subscriber get in touch at askresearch@hsbc.com