May was a game-changing month, not just in the world of AI but also for me personally. ABBYY hosted the AI Summit in London at the Emirates Stadium, and it was absolutely fantastic. One of the highlights for me was the fireside chat I hosted with industry experts on AI Privacy, Compliance, and Ethics. But you’ll be hearing all about that in next month’s edition.
Let’s focus on what’s happened in the AI “realm,” as LLMs love to say.
We’re seeing a market realization that AI isn’t all it’s cracked up to be—yet at the same time, the hype it’s experienced so far is actually not enough. What do I mean by cryptic message? Well, over the last month, several voices have advocated for more purposeful real-world usage of AI. To continue funding the discovery of the incredible things AI can do, we also need to find ways to make it truly useful for everyone—individuals and businesses alike.
It’s exactly that mix you’ll find highlighted in this month’s top AI news. , keep scrolling to get caught up with the latest developments from June.
Newsletter Special
Experience the new ABBYY Marketplace! It’s redesigned to kick start your automation initiatives easier and faster. Find purpose-built AI models that fuel LLMs and RAG integration. They’re ready to use for any document, no matter the format or language.
May’s big AI news revolves around the shift from flashy generative AI applications to practical, enterprise-focused solutions. WIRED highlights how startups like Tome and Perplexity are narrowing their focus to meet the specific needs of business clients, like sales and marketing teams. This pivot is driven by the high costs of AI development and the need for sustainable revenue streams. Enterprise applications may not have the same wow factor, but they promise real utility and value.
At ABBYY, we’ve been ahead of this curve with our Purpose-Built AI, delivering targeted solutions that address specific business challenges. Our customer stories showcase how tailored AI applications can drive real value, improve efficiency, and generate meaningful outcomes. It’s not always about the flashiest tech—sometimes, it’s about the solutions that quietly get the job done best. .
In the past six months, we've seen significant strides in AI transparency, but there's still a lot of work to do. The latest Foundation Model Transparency Index (FMTI) from Stanford reveals that developers are becoming more transparent, with scores improving from an average of 37 to 58 out of 100. This progress is promising, but areas like data access and the evaluation of model trustworthiness remain opaque.
It's just like navigating the layers of a dream in Inception—each layer reveals more, but there's always a deeper level to uncover. Just as Cobb and his team delve deeper into dreamscapes, we're exploring the complexities of AI transparency. The Index highlights both the advancements and the areas needing improvement, emphasizing the ongoing effort required to achieve full transparency. As AI continues to integrate into more aspects of society, ensuring clarity and accountability is more crucial than ever. Hit the link for the detailed findings to see how transparency in AI is evolving and what steps are still needed.
There was a fascinating discovery in the AI world with the research paper titled "Fishing for Magikarp," which tackles the issue of under-trained tokens in large language models. I’ll admit, I came for the Pokémon reference, but I stayed for the research!
Just like the elusive Magikarp in Pokémon, these glitch tokens can cause unexpected behaviors in AI, leading to errors or even exploitation. The study presents new methods to automatically detect these problematic tokens, aiming to improve the efficiency and safety of language models. While progress has been made, there's still a lot of work to be done in refining these systems to ensure they perform reliably.
ABBYY Spotlight
ABBYY Everest Group IDP PEAK Matrix 2024
Everest Group has named ABBYY a Leader in its Intelligent Document Processing (IDP) Products PEAK Matrix Assessment—for the 6th year in a row.
Discover how ABBYY’s Process Mining helped Emerson uncover critical insights, pinpoint the most efficient workflows and harness Process Simulation for informed decision-making.
Cancer Clinic Accelerates and Improves Patient Care with ABBYY IDP
Discover how a leading-edge, non-profit cancer treatment and research facility in Australia that provides compassionate and patient-centered care to over 60,000 people a year uses ABBYY’s intelligent document processing to accelerate access to medical records, reducing staff burnout, and allow more time to be spent on patients.
In May 1997, IBM's Deep Blue made history by defeating world chess champion Garry Kasparov in a six-game match. This event marked the first time a computer had beaten a world champion under standard chess tournament time controls, showcasing the incredible potential of AI in strategic thinking and problem-solving. The match not only elevated the public profile of AI but also demonstrated the significant advances in computing power and algorithmic sophistication.
You have received this e-mail because you are an ABBYY customer or partner, have requested further information on our products, solutions or services, registered for an ABBYY trial version, and/or agreed to receive news from ABBYY.
.png?upscale=true&width=166&upscale=true&name=newsletter-icons-02-removebg-preview%20(1).png){kind=icon}
