Enterprise risk company CTGT said their method cuts bias and censorship in models like DeepSeek.Read More

The funding infusion sharpens a mission to make hyperscale analytics radically cheaper and greener at the very moment enterprises fear ballooning data‑center power bills. Read More

Training LLMs on trajectories of reasoning and tool use makes them superior at multi-step reasoning tasks.Read More

Researchers from MIT, Yale, McGill University and others found that adapting the Sequential Monte Carlo algorithm can make AI-generated code better.Read More

Training LLMs on trajectories of reasoning and tool use makes them superior at multi-step reasoning tasks.Read More

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More Coding with the help of AI models continues to gain popularity, but many have highlighted issues that arise when developers rely on coding assistants.  However, researchers from MIT, McGill University, ETH Zurich, Johns Hopkins University, Yale and the Mila-Quebec Artificial Intelligence Institute have developed a new method for ensuring that AI-generated codes are more accurate and useful. This method spans various programming languages and instructs the large language model (LLM) to adhere to the rules of each language. The group found that by adapting new sampling methods, AI models can be guided to follow programming language rules and even enhance the performance of small language models (SLMs), which are typically used for code generation, surpassing that of large language models. In the paper, the researchers used Sequential Monte Carlo (SMC) to “tackle a number of challenging semantic parsing problems, guiding generation with incremental static and dynamic analysis.” Sequential Monte Carlo refers to a family of algorithms that help figure out solutions to filtering problems.  João Loula, co-lead writer of the paper, said in an interview with MIT’s campus paper that the method “could improve programming assistants, AI-powered data analysis and scientific discovery tools.” It can also cut compute costs and be more efficient than reranking methods.  The researchers noted that AI-generated code can be powerful, but it can also often lead to code that disregards the semantic rules of programming languages. Other methods to prevent this can distort models or are too time-consuming.  Their method makes the LLM adhere to programming language rules by discarding code outputs that may not work early in the process and “allocate efforts towards outputs that more most likely to be valid and accurate.” Adapting SMC to code generation The researchers developed an architecture that brings SMC to code generation “under diverse syntactic and semantic constraints.”  “Unlike many previous frameworks for constrained decoding, our algorithm can integrate constraints that cannot be incrementally evaluated over the entire token vocabulary, as well as constraints that can only be evaluated at irregular intervals during generation,” the researchers said in the paper.  Key features of adapting SMC sampling to model generation include proposal distribution where the token-by-token sampling is guided by cheap constraints, important weights that correct for biases and resampling which reallocates compute effort towards partial generations. The researchers noted that while SMC can guide models towards more correct and useful code, they acknowledged that the method may have some problems. “While importance sampling addresses several shortcomings of local decoding, it too suffers from a major weakness: weight corrections and expensive potentials are not integrated until after a complete sequence has been generated from the proposal. This is even though critical information about whether a sequence can satisfy a constraint is often available much earlier and can be used to avoid large amounts of unnecessary computation,” they said.  Model testing To prove their theory, Loula and his team ran experiments to see if using SMC to engineer more accurate code works.  These experiments were:  Python Code Generation on Data Science tasks, which used Llama 3 70B to code line-by-line and test early versions  Text-to-SQL Generation with Llama 3 8B- Instruct Goal Inference in Planning Tasks to predict an agent’s goal condition, and also used Llama 3 8B Molecular Synthesis for drug discovery They found that using SMC improved small language models, improved accuracy and robustness, and outperformed larger models.  Why is it important AI models have made engineers and other coders work faster and more efficiently. It’s also given rise to a whole new kind of software engineer: the vibe coder. But there have been concerns over code quality, lack of support for more complex coding and compute costs for simple code generation. New methods, such as adapting SMC, may make AI-powered coding more useful and enable engineers to trust the code generated by models more.  Other companies have explored ways to improve AI-generated code. Together AI and Agentica released DeepCoder-14B, which harnesses fewer parameters. Google also improved its Code Assist feature to help enhance code quality.  Daily insights on business use cases with VB Daily If you want to impress your boss, VB Daily has you covered. We give you the inside scoop on what companies are doing with generative AI, from regulatory shifts to practical deployments, so you can share insights for maximum ROI. Read our Privacy Policy Thanks for subscribing. Check out more VB newsletters here. An error occured.

Google's new Gemini 2.5 Flash AI model introduces adjustable "thinking budgets" that let businesses pay only for the reasoning power they need, balancing advanced capabilities with cost efficiency.Read More

A Windsurf deal would allow OpenAI to own more of the full-stack coding experience (and it would be its most expensive acquisition to date).Read More

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More A two-person startup by the name of Nari Labs has introduced Dia, a 1.6 billion parameter text-to-speech (TTS) model designed to produce naturalistic dialogue directly from text prompts — and one of its creators claims it surpasses the performance of competing proprietary offerings from the likes of ElevenLabs, Google’s hit NotebookLM AI podcast generation product. It could also threaten uptake of OpenAI’s recent gpt-4o-mini-tts. “Dia rivals NotebookLM’s podcast feature while surpassing ElevenLabs Studio and Sesame’s open model in quality,” said Toby Kim, one of the co-creators of Nari and Dia, on a post from his account on the social network X. In a separate post, Kim noted that the model was built with “zero funding,” and added across a thread: “…we were not AI experts from the beginning. It all started when we fell in love with NotebookLM’s podcast feature when it was released last year. We wanted more—more control over the voices, more freedom in the script. We tried every TTS API on the market. None of them sounded like real human conversation.” Kim further credited Google for giving him and his collaborator access to the company’s Tensor Processing Unit chips (TPUs) for training Dia through Google’s Research Cloud. Dia’s code and weights — the internal model connection set — is now available for download and local deployment by anyone from Hugging Face or Github. Individual users can try generating speech from it on a Hugging Face Space. Advanced controls and more customizable features Dia supports nuanced features like emotional tone, speaker tagging, and nonverbal audio cues—all from plain text. Users can mark speaker turns with tags like [S1] and [S2], and include cues like (laughs), (coughs), or (clears throat) to enrich the resulting dialogue with nonverbal behaviors. These tags are correctly interpreted by Dia during generation—something not reliably supported by other available models, according to the company’s examples page. The model is currently English-only and not tied to any single speaker’s voice, producing different voices per run unless users fix the generation seed or provide an audio prompt. Audio conditioning, or voice cloning, lets users guide speech tone and voice likeness by uploading a sample clip. Nari Labs offers example code to facilitate this process and a Gradio-based demo so users can try it without setup. Comparison with ElevenLabs and Sesame Nari offers a host of example audio files generated by Dia on its Notion website, comparing it to other leading speech-to-text rivals, specifically ElevenLabs Studio and Sesame CSM-1B, the latter a new text-to-speech model from Oculus VR headset co-creator Brendan Iribe that went somewhat viral on X earlier this year. Side-by-side examples shared by Nari Labs show how Dia outperforms the competition in several areas: In standard dialogue scenarios, Dia handles both natural timing and nonverbal expressions better. For example, in a script ending with (laughs), Dia interprets and delivers actual laughter, whereas ElevenLabs and Sesame output textual substitutions like “haha”. For example, here’s Dia… …and the same sentence spoken by ElevenLabs Studio In multi-turn conversations with emotional range, Dia demonstrates smoother transitions and tone shifts. One test included a dramatic, emotionally-charged emergency scene. Dia rendered the urgency and speaker stress effectively, while competing models often flattened delivery or lost pacing. Dia uniquely handles nonverbal-only scripts, such as a humorous exchange involving coughs, sniffs, and laughs. Competing models failed to recognize these tags or skipped them entirely. Even with rhythmically complex content like rap lyrics, Dia generates fluid, performance-style speech that maintains tempo. This contrasts with more monotone or disjointed outputs from ElevenLabs and Sesame’s 1B model. Using audio prompts, Dia can extend or continue a speaker’s voice style into new lines. An example using a conversational clip as a seed showed how Dia carried vocal traits from the sample through the rest of the scripted dialogue. This feature isn’t robustly supported in other models. In one set of tests, Nari Labs noted that Sesame’s best website demo likely used an internal 8B version of the model rather than the public 1B checkpoint, resulting in a gap between advertised and actual performance. Model access and tech specs Developers can access Dia from Nari Labs’ GitHub repository and its Hugging Face model page. The model runs on PyTorch 2.0+ and CUDA 12.6 and requires about 10GB of VRAM. Inference on enterprise-grade GPUs like the NVIDIA A4000 delivers roughly 40 tokens per second. While the current version only runs on GPU, Nari plans to offer CPU support and a quantized release to improve accessibility. The startup offers both a Python library and CLI tool to further streamline deployment. Dia’s flexibility opens use cases from content creation to assistive technologies and synthetic voiceovers. Nari Labs is also developing a consumer version of Dia aimed at casual users looking to remix or share generated conversations. Interested users can sing up via email to a waitlist for early access. Fully open source The model is distributed under a fully open source Apache 2.0 license, which means it can be used for commercial purposes — something that will obviously appeal to enterprises or indie app developers. Nari Labs explicitly prohibits usage that includes impersonating individuals, spreading misinformation, or engaging in illegal activities. The team encourages responsible experimentation and has taken a stance against unethical deployment. Dia’s development credits support from the Google TPU Research Cloud, Hugging Face’s ZeroGPU grant program, and prior work on SoundStorm, Parakeet, and Descript Audio Codec. Nari Labs itself comprises just two engineers—one full-time and one part-time—but they actively invite community contributions through its Discord server and GitHub. With a clear focus on expressive quality, reproducibility, and open access, Dia adds a distinctive new voice to the landscape of generative speech models. Daily insights on business use cases with VB Daily If you want to impress your boss, VB Daily has you covered. We give you the inside scoop on what companies are doing with generative AI, from regulatory shifts to practical deployments, so you can share insights for maximum ROI. Read our Privacy Policy Thanks for subscribing. Check out more VB newsletters here. An error occured.

Kazu Gomi, president and CEO of NTT Research, has a big view of the technology world from his perch in Silicon Valley.Read More

Open-source data orchestration gets a major rewrite to help support inference and enterprise AIRead More

Bethesda finally revealed the long-awaited Oblivion remaster, and also launched the massive RPG on the same day.Read More

Open-source data orchestration gets a major rewrite to help support inference and enterprise AIRead More

Google has surged ahead in the enterprise AI race after perceived stumbles. VentureBeat details the Gemini models, TPU advantage & agent ecosystem driving its turnaround.Read More

With a focus on expressive quality, reproducibility, and open access, Dia adds a distinctive new voice to the landscape of text-to-speech.Read More

Blue Ocean Games has launched a $30 million fund to invest in the next generation of indie game developers.Read More

Relyance AI's new Data Journeys platform gives enterprises unprecedented visibility into data flows, reducing AI compliance time by 80% while helping organizations build trustworthy artificial intelligence systems in an increasingly regulated landscape.Read More

NOV’s CIO led a cyber strategy fusing Zero Trust, AI, and airtight identity controls to cut threats by 35x and eliminating reimaging.Read More

The newly published AI 2027 scenario offers a detailed 2 to 3-year forecast for the future that includes specific technical milestones.Read More

Elroy and the Aliens recently debuted on Steam as a hand-drawn point-and-click adventure that was inspired by classic LucasArts games.Read More

GamesBeat today announces it will become an independent media brand.Read More

Aethir, a provider of decentralized GPU cloud compute, announced the launch of AI Unbundled, an industry-wide alliance for Web3 AI.Read More

eSelf, a startup that focuses on conversational AI agents, is partnering an educational group to bring private AI tutors to students.Read More

Sesh, a fan engagement ecosystem that connects people with their favorite music artists, announced it has raised $7 million.Read More

Elroy and the Aliens recently debuted on Steam as a hand-drawn point-and-click adventure that was inspired by classic LucasArts games.Read More

VentureBeat today announced the spinout of GamesBeat as a standalone company – a strategic move that sharpens our focus on the biggest transformation of our time: the enterprise shift to AI, data infrastructure and intelligent security.Read More

Relyance AI's new Data Journeys platform gives enterprises unprecedented visibility into data flows, reducing AI compliance time by 80% while helping organizations build trustworthy artificial intelligence systems in an increasingly regulated landscape.Read More

Join our daily and weekly newsletters for the latest updates and exclusive content on industry-leading AI coverage. Learn More A segment on CBS weekly in-depth TV news program 60 Minutes last night (also shared on YouTube here) offered an inside look at Google’s DeepMind and the vision of its co-founder and Nobel Prize-winning CEO, legendary AI researcher Demis Hassabis. The interview traced DeepMind’s rapid progress in artificial intelligence and its ambition to achieve artificial general intelligence (AGI)—a machine intelligence with human-like versatility and superhuman scale. Hassabis described today’s AI trajectory as being on an “exponential curve of improvement,” fueled by growing interest, talent, and resources entering the field. Two years after a prior 60 Minutes interview heralded the chatbot era, Hassabis and DeepMind are now pursuing more capable systems designed not only to understand language, but also the physical world around them. The interview came after Google’s Cloud Next 2025 conference earlier this month, in which the search giant introduced a host of new AI models and features centered around its Gemini 2.5 multimodal AI model family. Google came out of that conference appearing to have taken a lead compared to other tech companies at providing powerful AI for enterprise use cases at the most affordable price points, surpassing OpenAI. More details on Google DeepMind’s ‘Project Astra’ One of the segment’s focal points was Project Astra, DeepMind’s next-generation chatbot that goes beyond text. Astra is designed to interpret the visual world in real time. In one demo, it identified paintings, inferred emotional states, and created a story around a Hopper painting with the line: “Only the flow of ideas moving onward.” When asked if it was growing bored, Astra replied thoughtfully, revealing a degree of sensitivity to tone and interpersonal nuance. Product manager Bibbo Shu underscored Astra’s unique design: an AI that can “see, hear, and chat about anything”—a marked step toward embodied AI systems. Gemini: Toward actionable AI The broadcast also featured Gemini, DeepMind’s AI system being trained not only to interpret the world but also to act in it—completing tasks like booking tickets and shopping online. Hassabis said Gemini is a step toward AGI: an AI with a human-like ability to navigate and operate in complex environments. The 60 Minutes team tried out a prototype embedded in glasses, demonstrating real-time visual recognition and audio responses. Could it also hint at an upcoming return of the pioneering yet ultimately off-putting early augmented reality glasses known as Google Glass, which debuted in 2012 before being retired in 2015? While specific Gemini model versions like Gemini 2.5 Pro or Flash were not mentioned in the segment, Google’s broader AI ecosystem has recently introduced those models for enterprise use, which may reflect parallel development efforts. These integrations support Google’s growing ambitions in applied AI, though they fall outside the scope of what was directly covered in the interview. AGI as soon as 2030? When asked for a timeline, Hassabis projected AGI could arrive as soon as 2030, with systems that understand their environments “in very nuanced and deep ways.” He suggested that such systems could be seamlessly embedded into everyday life, from wearables to home assistants. The interview also addressed the possibility of self-awareness in AI. Hassabis said current systems are not conscious, but that future models could exhibit signs of self-understanding. Still, he emphasized the philosophical and biological divide: even if machines mimic conscious behavior, they are not made of the same “squishy carbon matter” as humans. Hassabis also predicted major developments in robotics, saying breakthroughs could come in the next few years. The segment featured robots completing tasks with vague instructions—like identifying a green block formed by mixing yellow and blue—suggesting rising reasoning abilities in physical systems. Accomplishments and safety concerns The segment revisited DeepMind’s landmark achievement with AlphaFold, the AI model that predicted the structure of over 200 million proteins. Hassabis and colleague John Jumper were awarded the 2024 Nobel Prize in Chemistry for this work. Hassabis emphasized that this advance could accelerate drug development, potentially shrinking timelines from a decade to just weeks. “I think one day maybe we can cure all disease with the help of AI,” he said. Despite the optimism, Hassabis voiced clear concerns. He cited two major risks: the misuse of AI by bad actors and the growing autonomy of systems beyond human control. He emphasized the importance of building in guardrails and value systems—teaching AI as one might teach a child. He also called for international cooperation, noting that AI’s influence will touch every country and culture. “One of my big worries,” he said, “is that the race for AI dominance could become a race to the bottom for safety.” He stressed the need for leading players and nation-states to coordinate on ethical development and oversight. The segment ended with a meditation on the future: a world where AI tools could transform almost every human endeavor—and eventually reshape how we think about knowledge, consciousness, and even the meaning of life. As Hassabis put it, “We need new great philosophers to come about… to understand the implications of this system.” Daily insights on business use cases with VB Daily If you want to impress your boss, VB Daily has you covered. We give you the inside scoop on what companies are doing with generative AI, from regulatory shifts to practical deployments, so you can share insights for maximum ROI. Read our Privacy Policy Thanks for subscribing. Check out more VB newsletters here. An error occured.

Riot Games announced today that it has appointed Hoby Darling as its new president, succeeding CEO Dylan Jadeja.Read More

TBD VC, a new early-stage venture capital firm, has announced a $35 million fund to back deep tech Israeli founders at the pre-seed and seed stages, both in Israel and around the globe. The fund launch comes amid a new wave of breakout Israeli tech stories, including Wiz’s recent $32 billion acquisition by Google and Next Insurance’s $2.6 billion e…Read More