The Discussion: Generative AI - from Consumer Rights to Human Rights #19
Jun 21, 2023
auto_awesome
Josh Muncke, Commercial Director at Faculty, and Sam Gregory, Executive Director at WITNESS, discuss generative AI models, prompt engineering, shallow fakes versus deep fakes, and the future of GPT. They explore the rapid progress of generative AI technology, its impact on customer experience, and the importance of human insight. They also touch on AI collaboration, risks and safeguards of language models, and the significance of media forensics.
Generative AI models are transforming the customer experience by creating personalized and unique interactions in retail and consumer industries.
The future of AI models involves optimizing prompts based on vectors, improving search results and customer responses.
The rise of deepfakes raises concerns about the potential risks and need for regulation, oversight, and careful consideration of use cases.
Deep dives
The rapid pace of AI technology
The podcast episode discusses the unexpected and rapid growth of AI technology, particularly in the last six months. The popularity and awareness of generative models, such as GPT, has increased exponentially, catching many experts off guard. This progress can be attributed to factors like improved accessibility, user interfaces, and growing awareness among regulators and legislators.
The evolving market and competition
The podcast explores the current landscape of AI technologies and their market. OpenAI, the company responsible for GPT, has emerged as a dominant player due to its advanced capabilities and focus on safety. However, the market is highly competitive, with other companies like Anthropic and Google making significant advancements. The market is volatile, and the future remains uncertain, with possibilities of consolidation or a tiered market for various specialized applications.
Applying AI in retail and consumer experiences
The podcast delves into the transformative potential of generative models in retail and consumer industries. These models offer the ability to create personalized and unique customer experiences. By utilizing AI in areas like customer service, marketing, and product visualization, retailers can augment their existing teams and improve efficiency and responsiveness. The possibilities extend to bespoke design choices, where personalized products can be generated based on individual preferences. Personalization becomes a key factor in customer engagement and delivering tailored experiences.
The Advancement of Generative AI and Language Models
Generative AI and language models are rapidly advancing, allowing for the optimization of prompts to generate high-quality outputs. The potential future involves prompts being based on vectors in an embedding space rather than traditional human language, resulting in improved search results and customer responses. As these models get integrated into education and various professions, people will need to develop the skill of co-creation and co-collaboration with AI models in order to work effectively with them.
Concerns about Deepfakes and the Need for Risk Management
Deepfakes, which involve the manipulation of images and videos, are a cause for concern. While face-swapping deepfakes are well-known, there are other types of manipulations happening as well, such as animation of photos and puppetry. The accessibility of tools for creating deepfakes raises risks in areas like risk management and customer service. Incorrect or manipulated information can lead to safety risks or commercial ramifications. It is important to carefully consider the use cases, regulate the technology, and implement processes and human oversight to ensure accuracy, quality, and mitigate potential risks.
In this episode of The Discussion I’m joined by Josh Muncke - Commercial Director, Retail & Consumer at Faculty and Sam Gregory - Executive Director at Human Rights Organisation, WITNESS.
We cover generative AI models, what they are and how they're changing the customer experience. Testing and iterating as a process, and how the field of prompt engineering is going to evolve. Also shallow fakes versus deep fakes, the novel ability to create new media content from existing data, the need for safeguards and transparency. And of course, what's next for GPT, and the next generation of models.
