6 Key Insights from Web Summit 2024 Worth Knowing

Tech leaders and innovators descended on Lisbon two weeks ago for Web Summit, Europe’s biggest tech conference to discuss the state of the industry and the trends that will be shaping it tomorrow. Locaria’s Thorsten Brueckner, senior data scientist, team leader, and strategic advisor, and Misha Pimenov, EVP of creative content, were in attendance and gathered key insights from the conference to share with LBB. 

A few trends emerged, but the dominant one was, of course, AI. According to Thorsten and Misha, the initial cycle of development and innovation seems to have slowed down with focus shifting to the technology’s applications. Here’s what they had to say:

So Much AI

The summit’s most discussed topic was AI with many of the speakers pointing to the ongoing exploration of AI, and the ethical and legal complications involved. AI is becoming a fundamental component of many products, but there is no clear market leader. The future of AI models is likely to focus on smaller, specialised models rather than one-size-fits-all solutions.

Video Intelligence and Asset Management Solutions

A masterclass about video intelligence emphasised AI's role in summarising video content and improving semantic search and asset tagging capabilities as online video content continues to boom. Tagging elements in videos to enhance search efficiency is going to be particularly significant for large clients with huge asset volumes. There’s a lot of potential benefits to integrating such technology into existing asset management systems to significantly improve efficiency. 

The Evolution of AI from LLMs to Specialised Models

The AI landscape is evolving, and fine-tuning large language models (LLMs) on proprietary data is becoming increasingly important. However, the process is far from simple. It requires not only gathering, cleaning, and annotating vast amounts of data but also handling complex workflows and computational demands. Fine-tuning locally is almost impossible for most, as it demands high-performance hardware and expertise that aren’t easily accessible. 

Instead, cloud-based solutions are often the only viable option, but they come with their own set of challenges, including steep costs and intricate set-up requirements. During IBM’s masterclass, the speakers introduced InstructLab, highlighting the need to simplify  fine-tuning, make it more cost-effective, and make it open-source. Frameworks like InstructLab aim to address the growing demand for smaller, highly specialised models.

The keynote by Thomas Wolf of HuggingFace emphasised a similar shift away from massive, general-purpose models. He pointed to a different future, one that lies in smaller, task-specific models working collaboratively. As the development of large models slows, this approach would provide a practical, efficient path forward, focusing on innovation that meets real-world needs.

Marketing Innovations and Customer Experience

Tiffany Rolfe's presentation introduced the term ‘omnimodal’ to describe a more strategic approach to customer engagement to fight against the generic Ecommerce experiences dominating the market. She raised the question of brand promptability and whether global brands are ready for an environment dominated by AI-powered tech, noting the growing competition between traditional search engines and AI tools like ChatGPT. Dan Gardner also called user experience into question, criticising the current state of the internet and its UX, predicting that AI could significantly enhance customer journeys.

The Implications of AGI Development

The next step from AI is artificial general intelligence (AGI), a system capable of reasoning the way humans do, and therefore able to perform tasks independently across different domains. There are concerns around its development. A fear that AGI could surpass human intelligence and lead to unpredictable negative outcomes. These anxieties don’t detract from a whole host of positives that could come from AGI, such as alleviating labour burdens and improving care for the elderly through robotics. 

Is it really AI?

These days, every company feels the need to market ‘AI’ in their products to stay competitive in a landscape where it has become the dominant buzzword. But often, it’s more about marketing than actual AI. Features that were once labelled as automation and used widely by many are being repackaged as ‘AI’. Decades-old statistical methods and modernised machine learning algorithms are being rebranded under the same term, even though the underlying technology remains largely unchanged. This doesn’t make the products or features less impressive—often, they deliver real value. However, the overuse of the term ‘AI’ risks diluting its meaning, making it harder to distinguish between genuine innovation and simple rebranding.