OpenAI and Scarlett Johansson
OpenAI faces a legal challenge from actress Scarlett Johansson after Sky, its newly released voice chatbot, appears to sound very similar to the actress herself.
Johansson was approached in September 2023 by OpenAI founder Sam Altman, and after considering the offer, she rejected it. In OpenAI's defence, Altman claims that they used a voice actor before any approach was made.
The problem with OpenAI and other high-value start-ups is that they use a standard Silicon Valley playbook, where they ignore everything around them and focus only on designing, disrupting, and delivering.
Copyright and intellectual property (IP) are among the areas causing problems for developers of large language models (LLMs) and generative AI solutions. While there is an appetite for the services that start-ups are working on and delivering, no business plan is yet in place to pay copyright and IP owners.
Just last week Sony Music Group issued a statement warming ‘companies to stop training AI on its artists’ content. Read their statement and declaration of AI training opt-out here.
The Silicon Valley Playbook
This playbook, pushed by venture capital companies, has produced some of the world's most influential tech companies. It revolves around a systematic approach to designing, developing, and delivering products or services.
Some of the critical components include:
User-Centric Design
Empathy for Users: Understanding user needs and pain points is fundamental. Companies invest heavily in user research, utilising techniques like surveys, interviews, and usability testing.
Iterative Design: A product or service is never finished. Instead, an approach of continuous iteration is used where new features are added following testing and feedback. This iterative process ensures that products align closely with user expectations and needs.
Lean Startup Methodology
Minimum Viable Product (MVP): The focus is on building a basic version of the product with essential features to test hypotheses about user needs and market demand. Using an Agile design methodology, products or services go through discovery, alpha and beta processes before they are released and then iterated.
Validated Learning: Every iteration of the product is an experiment to gather user feedback and data. This information guides future development and reduces the risk of failure.
Agile Development
Sprints and Scrums: Development is broken into short cycles (sprints) with regular evaluation (scrums). This approach allows teams to adapt quickly to changes and continuously improve.
Cross-Functional Teams: Teams often include members from different disciplines (engineering, design, marketing) to foster collaboration and innovation.
Having worked with many product and service teams, I can vouch for the importance of having diverse teams, especially when they include strategic communicators who can see what needs to be communicated and when to drive product/service adoption and engagement.
Data-Driven Decision Making
Analytics and Metrics: Decisions are based on data from user behaviour, product usage, and market trends. Key performance indicators (KPIs) are regularly monitored to assess progress and success.
A/B Testing: Experimentation with different versions of a product or feature helps identify what works best, optimizing user experience and business outcomes.
Insight gathered from engagement and testing helps shape various narratives, not just for iterative cycles but also for engagement with funders and investors to ensure that they have transparency and assurance that their investment will deliver a return.
Rapid Prototyping and Deployment
Continuous Integration and Deployment (CI/CD): Automated testing and deployment pipelines allow for rapid release cycles, enabling frequent updates and improvements.
Cloud Services: Leveraging cloud infrastructure provides scalability and flexibility, essential for handling growth and fluctuating demand.
Growth Hacking
Marketing and Product Synergy: Innovative and unconventional marketing strategies, often integrated into the product itself, drive rapid user acquisition and engagement.
Virality and Network Effects: Features that encourage sharing and collaboration can exponentially increase user base and engagement.
While pushing the boundaries of the rules and standards is good, communicators within design teams need to be equally aware of the risk from colleagues' decisions. Managing this risk is paramount, especially if investors are risk-sensitive.
OpenAI's major backer is Microsoft, a publicly listed company that needs to be more mindful of its culture of moving fast and breaking things.
Scalability and Adaptability
Modular Architecture: Products are designed with scalability in mind, using modular components that can be easily expanded or modified.
Pivoting: Companies remain flexible and willing to pivot or shift strategy based on market feedback and new opportunities.
Culture of Innovation
Risk-Taking and Failure Acceptance: Unlike the UK and other European nations, the US and Silicon Valley, in particular, have a culture that encourages experimentation and tolerates failure as a learning experience that fosters innovation.
Talent and Diversity: Attracting top talent and promoting diverse perspectives enhances creativity and problem-solving capabilities.
OpenAI and copyright and intellectual property
In growing at the pace that it has, OpenAI has had to use large data sets to train its large language models. ChatGPT works because of the data that it has been given access to. We don't know about the data that it has used or the value or payments for usage to train its own LLM.
Only last month in April 2024, did YouTube CEO Neal Mohan highlighted the possibility of OpenAI having used 'over a million hours of YouTube videos' to train ChatGPT-4.
Creatives, such as actors, writers, musicians, designers, and many more, make an income from the copyright they own.
A compensation model needs to be agreed upon between creatives, the owners of their artistic work, and digital and technology companies that use these assets to secure user engagement with their own products or services.
In essence, to safeguard the value and revenue and to manage the reputation of tech companies, which is needed if product adoption is going to be protected, the following considerations need to be planned:
Identify and classify creative assets and the owners of these
Create a revenue model based on subscription, licensing and API access
Establish partnerships and contractual agreements
Establish a process of transparency and accountability for data
Establish communication and stakeholder engagement processes within technology design teams
What we are currently seeing is a tug-of-war between copyright owners and technology disruptors.
Investors want to invest, but they also want to know that risks have been accounted for, managed and mitigated. If they have, then maybe even more investment can be given, knowing that there is clear and transparent revenue to be made in the future.
