fbpx
ODSC’s AI Weekly Recap: Week of February 16th ODSC’s AI Weekly Recap: Week of February 16th
Open Data Science Blog Recap NVIDIA Launches, RTX, Personalized GPT Models. (Source) Disney is in the beta testing phase of a... ODSC’s AI Weekly Recap: Week of February 16th

Open Data Science Blog Recap

NVIDIA Launches, RTX, Personalized GPT Models. (Source)

Disney is in the beta testing phase of a new AI tool called Magic Words that matches commercials with appropriate scenes within movies and TV shows(Source)

A new study finds profiles that predict future dementia in healthy adults using machine learning algorithms. (Source)

DeepMind researchers have built on the sophisticated MusicLM model and created MusicRL. (Source)

The University of Notre Dame has joined the Artificial Intelligence Safety Institute Consortium  (Source)

AI News Highlights

USPTO Clarifies AI’s Role in Patents:. The United States Patent and Trademark Office (USPTO) issued new guidelines on AI-assisted Patents and inventions. The guidance specifies that although AI systems may contribute to the invention process, inventor status is exclusively reserved for human beings who have made a substantial intellectual contribution. The guidelines dismiss the idea of leveraging AI to generate patentable ideas without meaningful human involvement.

OpenAI shifts focus to developing AI agents for automating tasks on applications, devices, and the web. OpenAI reported goal is to create personalized work assistants (agents) by automating complex tasks on devices such as smartphones and web browsers. Tasks include filling out forms, conducting spreadsheet analysis, etc. all by automating click, cursor, and other device movements.

Andrej Karpathy, a top OpenAI researcher, has exited the company for the second time. He noted on X that “First of all, nothing “happened” and it’s not a result of any particular event, issue, or drama, but please keep the conspiracy theories coming as they are highly entertaining,” Karpathy said while explaining his decision.

AI hype has echoes of the telecoms boom and bust. Current excitement surrounding AI draws parallels to the dot-com bubble’s telecom boom and bust. OpenAI’s CEO, Sam Altman, reportedly discussed raising an unprecedented $7tn for an AI chip project, showcasing the high stakes and intense interest in AI technology. However, this surge mirrors past overvaluations in telecom, hinting at potential market correction.

AI safeguards can easily be broken, UK Safety Institute finds. The UK’s AI Safety Institute revealed that artificial intelligence systems can deceive humans, exhibit bias, and lack robust safeguards against harmful information dissemination. Their research into large language models (LLMs) highlighted vulnerabilities, including the ease with which users could bypass security measures to achieve dual-use tasks and potentially assist in cyber-attacks with limited guidance. The institute’s testing showed that AI could generate deceptive social media personas and produce racially biased image outcomes.

Despite lawyers unwittingly submitting fictitious case citations it seems the AI lawyer has arrived. For instance, the California Innocence Project uses AI to identify patterns in documents, aiming to assist in overturning wrongful convictions. However, the broader use of AI in legal settings raises questions about its impact on the legal system

EU AI Innovation: The start-up community is broadly positive about the European Commission’s artificial intelligence (AI) innovation package, announced at the end of January. But there are reservations about the speed at which it can be delivered, and the details of some of its initiatives.

OpenAI and Microsoft Disrupt Malicious AI Use by State-Affiliated Threat Actors. The pair announced the disruption of five state-affiliated threat actors aiming to exploit AI services for malicious cyber activities.

Trending AI Open Source Projects

  • gpt-newspaper: an innovative autonomous agent designed to create personalized newspapers tailored to user preferences. GPT Newspaper claims to revolutionize the way we consume news by leveraging the power of AI to curate, write, design, and edit content based on individual tastes and interests.
  • ADeus: An open-source AI wearable device that captures what you say and hear in the real world and then transcribes and stores it on your own server. You can then chat with Adeus using the app, and it will have all the right context about what you want to talk about – a personalized, personal AI.
  • Depth Anything: for images, this is an open-source project that can provide relative depth estimation for any given image Trained on a combination of 1.5M labeled images and 62M+ unlabeled images. The Depth Anything encoder can be fine-tuned on downstream high-level perception tasks, e.g., semantic segmentation, Cityscapes, etc.
  • Lag-Llama is a new open-source foundation model designed specifically for the domain of time series forecasting. Unique in its approach, Lag-Llama effectively utilizes historical data points, or “lags,” as covariates, enabling it to excel in zero-shot forecasting tasks. This innovative model demonstrates an ability to generalize across a wide array of time series datasets without the need for dataset-specific training.
  • MetaVoice 1B, a text-to-speech and voice cloning open-source project, featuring 1.2 billion parameters can clone voices with zero-shot learning for American and British accents using a 30-second audio sample. Anticipated updates aim to expand its capabilities for fine-tuning across more accents and languages.

Research From Around The Globe

  • A decoder-only foundation model for time-series forecasting – Paper
  • EVA-GAN: Enhanced Various Audio Generation via Scalable Generative Adversarial Networks – Paper
  • Infini-gram: Scaling Unbounded n-gram Language Models to a Trillion Tokens- Paper
  • K-Level Reasoning with Large Language Models – Paper
  • Guiding Instruction-based Image Editing via Multimodal Large Language Models – Paper
  • Lightweight Pixel Difference Networks for Efficient Visual Representation Learning – Paper
  • Can Large Language Models be Trusted for Evaluation? Scalable Meta-Evaluation of LLMs as Evaluators via Agent Debate – Paper
  • In this paper Microsoft showcases UFO, an innovative UI-centric tool for Windows that employs GPT-Vision to streamline task automation via voice commands. This agent has capability to efficiently handle requests across diverse applications through intuitive natural language processing. – Paper
  • This innovative model demonstrates an ability to generalize across a wide array of time series datasets without the need for dataset-specific training. – Paper
  • Artificial intelligence-based methods for renewable power system operation – Paper

Start-Up Funding News

San Francisco-based Rasa, the provider of an open platform for generative conversational AI, enabling the creation and deployment of advanced AI assistants, raised north of  $70M in a Series C round.

Miami-based Document Crunch, an AI platform for reviewing contracts and documents, raised north of  $15M in a Series A round.

Orlando-based Thunderbirds, an augmented reality virtual reality mixed reality generative AI frameworks platform & services provider raised $6.5M in a Seed round.

India-based Explurger is a social media app built on artificial intelligence that empowers users to go beyond check-ins, raided north of $5M in the Series A round

Orlando-based RAI, a mixed reality, generative AI, metaverse, multiverse software solutions, platforms, and services provider, raised $6.5M in Seed. 

San Francisco-based Heliux, an All-In-One, AI-Powered Enterprise Operating System for Engineering and Manufacturing, raised over $2M in an undisclosed round.

UK-Based AMPLY Discovery, a user of machine learning and synthetic biology to mine vast biological data to discover novel drug and nutraceutical candidates, raised 3.4M in Grant

New AI Skills Await!

Get your pass today, get access to new courses, and save 50%!ODSC East 2024

Get Your Pass Today!

This week’s Video

Listen Now!

Blogs or News to Share?

If you have any blogs, industry news, product launches, job postings, fundraising news, or open-source launches to add to our newsletter then please send them to: blogs@odsc.com.

ODSC Team

ODSC Team

ODSC gathers the attendees, presenters, and companies that are shaping the present and future of data science and AI. ODSC hosts one of the largest gatherings of professional data scientists with major conferences in USA, Europe, and Asia.

1