BLINK Unpacked | The Year of GenAI: Technology’s Generational Moment

The Panel

Luca Romagnoli, RVP Global Growth Initiatives, Salesforce 
Andy Hood, VP Emerging Technologies, WPP
Nick Jakobi, Director of Product, Cohere
Anna Koivuniemi, Head of Google Deepmind, Google 
Miguel Martinez, Senior Deep Learning Data Scientist, NVIDIA

We recently hosted a panel of distinguished GenAI experts who shared their perspectives on how the space will evolve in the medium to long-term. We heard from leaders with leading perspectives on driving solutions within their  organisations, providing products for others, leading language modellers and the importance of communities working together around ethics.

Is GenAI hype or real value?

There was a consensus that GenAI has stirred significant discussion and conjecture in the past year. As we navigate a phase of increased economic unpredictability, corporate leaders are grappling with questions about where to cut costs and where to invest more; many have felt pressure to invest early in this emerging technology, but also don’t want to fall for empty hype. The collective agreement acknowledged that there won’t be an industry that GenAI does not impact and touch. Machine learning tools, like ChatGPT, have facilitated elevated levels of creativity and innovation and are already proving their value.

What is the advice for companies wanting to implement Generative AI?

Our panel discussed instances where their teams have quietly explored the potential of GenAI, seeking meaningful ways to enhance value. By investing in the development of internal tools, GenAI testing has yielded impressive results, garnering strong internal support within organisations. Concrete examples emerged, illustrating its ability to save time, boost productivity, and create more effective products for customers. As the world moves in this direction, having a comprehensive understanding of GenAI is imperative for all.

The adoption of GenAI at the enterprise level represents unchartered territory. Addressing board level concern about GenAI and the pressure to invest without clear guidance is a challenge. Some firms have brought in experts to showcase the current state of the technology, highlighting both its strengths and limitations.

GenAI is advancing rapidly compared to other technologies, with the panel agreeing that it is highly unlikely that employees are not using machine learning in some format. Due to this, setting a corporate strategy, guardrails and guidelines around GenAI is a necessity as is revisiting this at least twice a year due to the pace of change. Some firms are being more proactive, hiring specific researchers to help build preparatory models and solutions, offering an environment of experimentation and being courageous about making organisational change. Machine learning was considered ineffective without precise organisational data; hence, it is crucial for organisations to have a firm grasp on their data strategy. Talent was also discussed, with the panel believing that the organisations that will handle this period of uncertainty the best, will have leaders at the top of the organisation with deep technology expertise.

What is the future of generative AI and where is it going?

The panel delved into the future of AI, stressing that it is moving faster than any other technologies they have been involved with. It is still in an exponential growth period and using a high level of compute. Google has categorised artificial GenAI into five levels, ranging from level 0 (basic) to level 5 (superhuman). The consensus was that level 5 capability has not yet been achieved – “there is so much we don’t know yet”. The panel found the industry's pace intriguing and it is hard to know where it will stop.

The importance of multi-modality, advancing beyond voice and text, to accommodate multiple inputs was discussed, as was reliability and productivity improvement. The future of open-source models, such as LLaMA and Vicuna were also topics of interest. An example shared was around Google Deepmind’s AI tool GNoME recently discovering hundreds of thousands of new stable materials, equivalent to nearly 800 years’ worth of knowledge and the follow-on impact this could have on various industries including drug discovery.

It was agreed that the future of GenAI is a vast topic, however the overall sentiment was of excitement and positivity, emphasising how GenAI can add to people’s lives and professional capabilities, rather than take away.

Ethics and regulations?

GenAI has been applied across diverse fields, spanning art, content creation, drug discovery, and more. The panel explored ethical and legal considerations, addressing issues such as state-sponsored interference, watermarking AI-generated content, and the imperative of prioritising privacy. The consensus among our panel members was that international regulation has struggled to keep pace with the rapid changes in the field. To address these challenges, principles for model development have been established, and cross-functional teams are working to ensure responsible development.

The implementation of Red Team testing underscores the recognition that safety is subjective and varies among individuals. The panel stressed the necessity of building a more nuanced approach to ensure responsible development and deployment of GenAI. Several companies are investing in ethics and safety researchers who actively engage with experts and society, emphasising the imperative to use GenAI for the betterment of humanity.