Google’s updates in GenAI – key takeaways from Future of GenAI event in Finland

The Gen AI challenges, requirements for clear use cases and importance of responsibility and security were the key take aways in the event. As both CEO, Co-Founder Peter Sarlin from Silo AI and Founder, R&D Antero Hanhirova from Happeo opened a wide range of Gen AI related business benefits, which become realized when understanding how to command GenAI, prototyping different approaches and having a clear problem statement.

Google’s next generation Large Language Model PaLM2 and how that’s going to change how we work and do business

Google EMEA’s Generative AI Ambassador Ted Schönbeck’s keynote speech was eye opening when presenting Google’s responsible AI. PaLM2 is Google’s approach in building and deploying AI responsibly. It’s Google’s next generation large language model which is trained with much more data than its predecessor and has higher performance. You can also check out an explainer video if you are not familiar with the topic.

The new conversational AI platform BARD which is designed for everyone, is able to handle a massive dataset of text and code. It is not limited to a closed set of responses, so it’s quite accurate when processing the information from the real world in real time. Addition to code and scripts BARD is able to generate also, email, letters, etc. It can be used for a variety of purposes, from creative writing to technical documen­­tation and even to creating music, images, and videos.

Schönbeck was strongly emphasizing the importance of following EU’s data protection requirements what comes to the launch of BARD in Europe, platform needs to meet the compliance requirements before the launch. So, in addition to the existing 180 countries with three languages, BARD will be opened also in EU this year - and even in Finnish!

BARD is still under development, but it has the potential to revolutionize the way we interact with computers. It is a powerful new tool that can be used for a variety of purposes, and it is sure to have a major impact on the way we live and work for instance customer service, education, research, and creativity.

For the enterprise usage Google has launched Enterprise Generative AI. This platform is integrated end-to-end enterprise platform for business users, developers, and AI experts. The Enterprise Generative AI enables open and shared innovation in a secured way. Google Cloud Tenants are able to utilize the Model Garden library of more than 80 LLM’s, but

• Data is always encrypted,
• Google processes input only to generate the output,
• data & parameters are not stored, and
• all created innovation models and solutions stay inside the Cloud Tenant.

Google is aiming to generate 80% faster and 50% cheaper model training and tuning with The Enterprise Generative AI Platform’s performance, scale, and efficiency capabilities.

Utilizing GenAI comes also with AI responsibility

Google’s Technical Director Lara Suzuki gave an inspirational speech on AI responsibility. She considers that trusting AI systems is the biggest barrier for enterprises moving from the lab to production. Lara Suzuki keeps the ethics at the center of everything and considers only responsible AI being successful AI. She reminds that responsible AI is not only about the tooling (data exploration and visualization, model analysis and explainability methods and model reporting), it’s also very much about the corporations’ internal processes (governance such as approval processes, education, and AI development framework).

Suzuki’s guidance for all of us concerning responsible AI is very assuring – everyone has the same challenges: there is no one-size-fits-all approach, but definitions and standards are evolving and AI specific regulation is coming. Technologies as well as harms and risks are constantly changing. It’s also good to remember that national and cultural differences matter and in addition often culture change is required among sociotechnical approach.

Suzuki’s key steps to responsible AI are:

• Assess risk: define the use case, assessing impact, model limitations
• Consider adjustments: design and fine tuning, user safeguards, regulatory preparation
• Perform safety testing: safety attributes & thresholdings, adversials testing, testing datasets
• Model in production: model monitoring, user feedback, model cards, using RAI tooling

SAP and Google cooperation as a use case for the enterprise level organizations

Google’s SAP and Cortex Specialist Barrett Conway opened in his presentation the benefits of SAP and Google Cloud partnership launched this year. Customers are able to build an end-to-end data cloud that brings data from across the enterprise landscape utilizing SAP Datasphere and Google’s Data Cloud to view their entire data estates in real time and maximize value from their Google Cloud and SAP investments.

According to Barrett Conway data is the cornerstone of digital transformation and AI development, but organizations are struggling to get the value out of the data:

• 97% are investing in data initiatives
• 91% have increased investments in data and AI
• 74% appointed have chief data or analytics officers,
• but only 20% see their models go into production.

When combining SAP software data and non-SAP data on Google Cloud from any data source, organizations are able to run digital transformation with the full data foundation to utilize their full business potential: e.g. simplify integration complexity with SAP, reduce time, effort and cost to deliver analytics and insights and expertly assist with decision making and tasks.

SAP systems together with Google’s data cloud introduces entirely new opportunities for enterprises to derive more value from their full data footprints and unified data foundation and advanced AI tools and large language models to find new insights from their data. But it also addresses key sustainability challenges, Barrett Conway highlights.

SAP is able to unlock sustainability data in business processes while Google Cloud can unleash the power of sustainability data. The combination creates a Sustainability Data Framework which provides actionable insights to support corporate sustainability targets to prevent e.g. climate risks.

Google Cloud is a strategic partner for Tietoevry Tech Services and improving efficiency, sustainability, and innovation with us. We appreciate Google’s approach to Responsible AI, and we are eager to co-innovate with their advanced skills in generative AI which enables our customer’s success.