Let's have a chat!
May 2024 has brought significant advancements and regulatory milestones shaping the landscape of artificial intelligence. From OpenAI’s introduction of GPT-4o to Google’s updates at Google I/O 2024 and key developments in the European AI regulation, follow the latest innovations driving transformative AI applications across various sectors with additional insights from Martin Dostál, Look AI Ventures Investment Committee Member and AI expert.
The highlights of AI news for May 2024 include:
Earlier this May, OpenAI introduced GPT-4o, an upgraded version of its GPT-4 model. This advancement significantly enhances ChatGPT’s capabilities, supporting real-time spoken conversations and processing text, images, and more. With features like memory learning and multilingual translation, GPT-4o showcases its versatility.
The model’s new memory feature allows it to learn from previous interactions, providing a more personalized experience. Additionally, it features multilingual translation and emotion detection abilities. An interesting demonstration showed ChatGPT using its voice capabilities to solve math problems, narrate bedtime stories, and even sing. This release positions OpenAI ahead in the competitive AI landscape, with accessibility to unpaid users and integration into Microsoft’s products.
Martin Dostál notes that “GPT-4o makes a considerable move in the direction needed to address some existing drawbacks, such as more personalization and better multilingual translation. It also harvests the potential for technology advancements that we see in more extensive and better-implemented support of non-textual modalities such as images and voice.”
Google I/O 2024 showcased Google’s significant AI advancements across various fronts last month in California. Highlights included a new AI-powered search engine with expanded capabilities and the introduction of Gemini Live, which enables interactive voice chats with advanced speech and image recognition. The unveiling of Project Astra promises multimodal AI experiences, blending sight, sound, voice, and text interactions.
Google Workspace now incorporates AI helpers across apps like Gmail, Drive, and Docs. Additionally, Google introduced LearnLM for educational AI tutoring and Gemini, which enhances Gmail, Google Maps with generative AI summaries, and Android with deeper integration capabilities. Find out more about the 100 news Google announced in May.
Martin Dostál comments on these innovations, emphasizing that “both Gemini and Astra are very promising projects addressing an important and hot topic—productivity—from which every human can benefit.
“From a broader perspective, these advancements are truly exciting—augmented reality and natural multimodal interaction were first introduced as early working prototypes in the early 1980s. Systems such as “Put That There” were much less powerful, but the idea and concept remained. These days, technological advancements are making Google Gemini, Astra, or Microsoft Copilot materialize such a concept in a much more usable way.
One more critical note: Google needs to be more careful in validating their LLMs to limit the nonsense or dangerous outputs.”
Backed by Amazon and Google, Anthropic is launching its AI chatbot Claude in Europe, positioning it as a competitor to OpenAI’s ChatGPT.
Available via web and mobile app, Claude offers both free and paid subscription options, including a business plan at approximately $30 per month. Following its API launch earlier this year, Anthropic intends to seamlessly integrate Claude into workflows across European languages. Future enhancements may include image recognition capabilities. Emphasizing ethical AI, Anthropic highlights Claude’s reliability in generating safe content.
Regarding Anthropic’s entry into the European market, Martin Dostál expresses interest in Claude’s performance, stating, “I am very curious about the experience gained with Claude and whether the estimated benefits and weaknesses compared to ChatGPT will be true. It is widely estimated that Claude should better understand human language and provide more accurate responses, while ChatGPT seems to be better at creating content.”
In May 2024, Microsoft launched its Copilot+ PCs, a new category of Windows devices powered by advanced AI capabilities. These PCs, developed in collaboration with major manufacturers like HP, Dell, Samsung, and Lenovo, feature Qualcomm Snapdragon X series processors for high performance and extended battery life.
Promising to revolutionize the PC experience, Copilot+ integrates AI at every level—from hardware to software—enhancing tasks like file searching, email summarization, and smart scheduling. With robust security features and native support for popular apps, these devices are positioned to boost productivity and efficiency in enterprises. The launch signals a significant advancement in AI-driven computing tailored for both personal and professional use. The initial wave of Copilot+ PCs will be available for purchase from June 18.
Martin Dostál highlights the importance of vertical integration in leveraging AI potential: “Deep vertical integration is a must to fully leverage the potential of AI on end-user systems. Microsoft is very well aware of it and is well-positioned to do so as they are one of the leaders in AI.
“Apple did the same on the system level and the user interface and it is still one of its main competitive advantages—its products and ecosystem are tightly integrated. Not surprisingly, Apple now also wants to integrate AI into its ecosystem. Apple is better prepared for user interfaces and vertical integration, while Microsoft is better prepared for AI. However, the theory is one big caveat—privacy concerns, as a tight integration will need to have access to data and information. Balancing capabilities and privacy is a challenge.”
Google DeepMind has unveiled AlphaFold 3, an AI model capable of predicting the structures and interactions of a wide range of biological molecules, including proteins, DNA, RNA, and potential drug compounds. This advancement promises to revolutionize biological research by accelerating the understanding of molecular interactions, traditionally a costly and time-consuming process.
AlphaFold 3 extends the capabilities of its predecessors by offering state-of-the-art accuracy in predicting complex molecular structures and interactions, marking a significant leap toward computational solutions in drug discovery and biomedical research. This innovation is set to redefine the landscape of scientific inquiry and pharmaceutical development globally.
Reflecting on the broader implications of this breakthrough, Martin Dostál remarks, “Medicine and healthcare are some of AI’s most promising application domains. Interestingly, while Generative AI is currently at a prime time, the application of AI for generating new structures and modeling interactions is also a kind of generative AI that has been here for years already, and before the rise of Generative AI in the area of generating images or audio.”
On May 21, the European Council approved the first Artificial Intelligence Act, marking a significant global milestone in AI regulation. This new legislation adopts a risk-based approach, imposing stricter rules for AI systems of higher societal risks.
The EU AI Act aims to ensure safe and trustworthy AI deployment across the EU, balancing innovation with fundamental rights protection. Prohibited practices include AI for social scoring and biometric categorization based on sensitive attributes. Governance bodies and penalties are established to enforce compliance, while measures support innovation through regulatory sandboxes. Effective in two years, the act sets a global precedent, emphasizing transparency, accountability, and ethical AI development.
Martin Dostál underscores the broad impact of this legislation: “The EU AI Act will affect all providing or deploying AI solutions, including startups. All involved parties must prepare for the requirements and obligations resulting from it.
“It might be good to summarize which are the main steps that need to be taken:
“High-risk AI systems pose a high risk to citizens’ health, safety, and fundamental rights. For example, a large subset of healthcare applications, such as medical systems purported for diagnosis, monitoring physiological processes, and treatment decision-making, will typically be classified as high-risk systems. The EU AI Act defines two categories of so-called low risk: limited risk and minimal risk. A typical example of limited risk systems will be chatbots that do not fall into high-risk or unacceptable risk categories (e.g., a chatbot for medical diagnosis).”
Crunchbase reports that global venture funding reached $31B in May 2024, marking the highest monthly total this year and a notable increase of 40% from the previous month. AI secured the largest share of investments, totaling $12.5B across over 250 companies.
The month also saw a wave of billion-dollar rounds, with $11B raised by six companies, including xAI, CoreWeave, and Wayve. Healthcare and biotech followed with $5.1B in funding.
Elon Musk’s AI startup, xAI, has raised $6B in funding, valuing the company at $18B. Founded in 2023, xAI focuses on developing large language models crucial for generative AI advancements. This funding round, among the largest in AI, aims to bring xAI’s products to market, enhance infrastructure, and advance R&D.
Investors include Tesla backers like Andreessen Horowitz, Sequoia Capital, Fidelity Management & Research Company, and Kingdom Holding, a Saudi investor run by Prince Alwaleed bin Talal, a member of the Saudi royal family. Musk’s move into AI intensifies competition with OpenAI, which he previously co-founded.