Is It Possible for AI to Advance Our Understanding of the Natural World?

In his welcoming remarks, National Academy of Medicine President Victor Dzau emphasized that AI is helping science in many ways, including by predicting outcomes from data, simulating complicated scenarios, and finding relevant trends in huge datasets.

According to a report in the Nature journal on December 21st, researchers from Carnegie Mellon University made history by creating the first ever non-organic intelligent system to conceive, plan, and carry out a chemistry experiment.

Read:The Top AiThority Articles Of 2023

Features

Assistant Professor of Chemistry and Chemical Engineering Gabe Gomes, along with Daniil Boiko and Robert MacKnight, who were doctoral students in chemical engineering, created the system known as Coscientist. The complete spectrum of the experimental procedure is executed with a simple, plain English prompt using large language models (LLMs), such as OpenAI’s GPT-4 and Anthropic’s Claude.

The research team proved in their Nature paper that Coscientist can do things like plan the chemical synthesis of known compounds, navigate hardware documentation, control liquid handling instruments, complete scientific tasks involving multiple hardware modules and diverse data sources, and solve optimization problems by analyzing previously collected data. They also used the documentation to execute high-level commands in an automated lab called a cloud lab.

Read: State Of AI In 2024 In The Top 5 Industries

Cloud Lab

In early 2024, the first academic cloud lab will be opened by Carnegie Mellon in collaboration with ECL. More than 200 pieces of equipment will be available to academics and collaborators at Carnegie Mellon University through the Cloud Lab. Gomes intends to keep working on the technologies detailed in the Nature article so that they can be utilized by the Carnegie Mellon Cloud Lab and other autonomous research facilities down the road.

Additionally, a coscientist essentially unlocks the “black box” of research. By tracking and documenting each stage of the research process, the system ensures that the work can be easily reproduced.

The post Artificially Intelligent ‘Coscientist’ Automates Scientific Discovery appeared first on AiThority.

LLM in a Flash

Published on December 12, the new study titled “LLM in a Flash: Efficient Large Language Model Inference with Limited Memory” can revolutionize the iPhone experience. It could bring a more immersive visual experience and make complex AI systems accessible on iOS devices. A recent research paper by Apple reveals a groundbreaking method that can assist in implementing AI on iPhones.

New: 10 AI ML In Personal Healthcare Trends To Look Out For In 2024

3D animated avatars from single-camera footage

Apple researchers present HUGS (Human Gaussian Splats) as a method to create 3D animated avatars from single-camera footage in the first study. In a statement made by principal author Muhammed Kocabas, the researchers claimed that their system could automatically separate a static scene from an animated human avatar in as little as 30 minutes using only a monocular video with a modest number of frames (50-100).

Read: Top 10 Benefits Of AI In The Real Estate Industry

Future demands of AI-infused services

Apple is looking ahead to the future demands of AI-infused services as it considers incorporating these breakthroughs into its product selection, which might improve its gadgets even further. If Apple’s new memory-allocation feature works as advertised, it might pave the way for a whole new category of apps and services to take advantage of LLMs in ways that weren’t possible before.

In addition, Apple is contributing to the larger AI community by publicizing its research, which could encourage other improvements in the field. That Apple is willing to do this shows how seriously it takes its role as a technological leader and its dedication to expanding human potential.

Read:The Top AiThority Articles Of 2023

Flash storage optimization

Using flash storage optimization, this method streamlines large LLMs. Another major development will occur when Apple incorporates sophisticated AI inside the iPhone. Two new research papers showcased this month by the Cupertino-based tech behemoth declared substantial advancements in AI. The study uncovered novel methods for efficient inference of language models and 3D avatars. This research delves into the difficulty of keeping model parameters in flash memory, running them into DRAM on demand, and executing LLMs that use more DRAM than is available. Data transfers from flash memory can be optimized with the use of the Inference Cost Model, which takes flash and DRAM characteristics into account.

To back up their claim, the researchers have utilized models like Falcon 7B and OPT 6.7B. According to the research, compared to conventional approaches, the models increased CPU speed by 4-5 times and GPU speed by 20-25 times.

Read: State Of AI In 2024 In The Top 5 Industries

Why the users should be happy?

Users of Apple products, such as the iPhone, may profit substantially from the results of the study on efficient LLM inference with limited memory. Users will get access to greater AI capabilities with strong LLMs running efficiently on devices with limited DRAM, like as iPhones and iPads. Better language processing, smarter voice assistants, better privacy, maybe less internet bandwidth utilization, and, most significantly, making advanced AI available and responsive to every iPhone user—these are all features that come with the iPhone.

The post Revolutionary AI Breakthrough: Experience an Unparalleled Transformation of Your iPhone with Apple’s Latest Research appeared first on AiThority.

The post How to Incorporate Generative AI Into Your Marketing Technology Stack appeared first on AiThority.

How is AI helping Researchers Identify a New Class of Antibiotic?

To combat diseases caused by bacteria that are resistant to many antibiotics, artificial intelligence has been essential in the discovery of a new class of medications. This may be useful in the fight against antibiotic resistance, which is a growing problem that killed over 1.2 million people in 2019 and will likely continue to do so for decades to come. A novel antibiotic that can kill a type of bacterium responsible for many drug-resistant diseases has been identified by researchers at MIT and McMaster University using an artificial intelligence algorithm.

Read:The Top AiThority Articles Of 2023

The medicine has the potential to battle Acinetobacter baumannii, a type of bacteria commonly found in healthcare facilities and a cause of pneumonia, meningitis, and other severe diseases if it were to be developed for use in patients. Wounds sustained by troops serving in Iraq and Afghanistan are also frequently infected with this particular bacterium. Using a machine-learning model they trained to determine if a chemical compound inhibits the growth of A. baumannii, the researchers were able to identify the novel medicine from a library of roughly seven thousand potential medicinal molecules.

Read: State Of AI In 2024 In The Top 5 Industries

What are its features?

While very few new antibiotics have been created during the past several decades, many pathogenic bacteria have grown progressively resistant to current ones.

Collins, Stokes, and Regina Barzilay, a professor at MIT and co-author of the current paper, set out a few years ago to tackle this increasing problem using machine learning, an AI technique that can learn to identify patterns in massive datasets. Collaborating with MIT’s Abdul Latif Jameel Clinic for Machine Learning in Health, Collins and Barzilay intended to find novel medicines with structurally distinct molecular bonds using this method.

New: 10 AI ML In Personal Healthcare Trends To Look Out For In 2024

First, they showed that they could train a machine-learning system to find chemical compounds that could stop E. coli from growing. The researchers named the molecule halicin after the fictitious AI system from “2001: A Space Odyssey.” The algorithm produced it from a screen of over 100 million molecules. In addition to killing E. coli, they demonstrated that this chemical might eradicate other treatment-resistant bacterial species. After training the algorithm, scientists fed it data from the Broad Institute’s Drug Repurposing Hub, which included 6,680 novel molecules. A few hundred high-quality results were produced by this analysis, which did not take more than two hours. Researchers focused on compounds with structures different from current antibiotics or molecules from the training data, choosing 240 to test experimentally in the lab.

Read: Top 10 Benefits Of AI In The Real Estate Industry

A New Class of Antibiotic Candidates That Can Kill a Drug-Resistant Bacterium

Nine antibiotics, including a highly effective one, were produced during those tests. This chemical, which was first investigated for its use as a diabetes medication, was found to be highly efficient against A. baumannii but inactive against other bacterial species such as Pseudomonas aeruginosa, Staphylococcus aureus, and carbapenem-resistant Enterobacteriaceae.

Antibiotics are highly prized for their “narrow spectrum” killing capabilities, which reduces the likelihood of bacteria quickly developing resistance to the medicine. A further perk is that the medicine will probably not harm the good bacteria already present in the human digestive tract, which helps to prevent opportunistic illnesses like Clostridium difficile.David Braley Center for Antibiotic Discovery, Weston Family Foundation, Audacious Project, C3.ai Digital Transformation Institute, Abdul Latif Jameel Clinic for Machine Learning in Health, DARPA Accelerated Molecular Discovery, Canadian Institutes of Health Research, Genome Canada, McMaster University’s Faculty of Health Sciences, Boris Family, a Marshall Scholarship, and the Department of Energy Biological and Environmental Research program were among the organizations that contributed to the funding of this research.

The post Using AI, Researchers Identify a New Class of Antibiotic Candidates That Can Kill a Drug-Resistant Bacterium appeared first on AiThority.

Lenovo, a global tech powerhouse, takes center stage in 2023 with a cascade of exciting news stories that underscore its continued influence in the ever-evolving world of technology. Embarking on the journey of a new year, Lenovo sets the tone with a series of top-tier developments, positioning itself at the forefront of innovation and progress in 2023.

In the fast-paced landscape of consumer electronics, Lenovo’s top 10 news stories for 2023 unfold as a tapestry of advancements, collaborations, and strategic moves, reflecting the company’s commitment to staying ahead in the competitive tech market. As the digital era progresses, Lenovo showcases its prowess with a collection of transformative narratives, ranging from cutting-edge product launches to strategic partnerships that redefine the boundaries of technological possibilities.

Lenovo’s 2023 story is one of resilience and innovation, where each news headline paints a picture of a company determined to shape the future of computing and redefine the user experience.

Top 10 News of Lenovo in 2023

Lenovo Introduces Feature-Packed Devices and Solutions at CES that Deliver Users a More Personalized Experience

Lenovo Launches New Hybrid Cloud Platforms and Services to Accelerate AI

Lenovo announced it has expanded its hybrid cloud platform for AI with new ThinkAgile hyperconverged solutions and ThinkSystem servers that advance cloud deployment, hybrid connectivity and AI capabilities, powered by the next generation of Intel® Xeon® Scalable Processors. The new AI ready platform delivers improved performance and the latest accelerators as a critical next step for delivering a dynamic hybrid AI approach across public, private, and foundational models to enable AI for All.

The new Lenovo ThinkAgile hybrid cloud solutions are engineered to boost AI performance and build cloud agility by delivering more compute and faster memory to its market-leading portfolio when and where it is needed. Additionally, Lenovo Professional Services for AI and TruScale as-a-service offerings help customers simplify IT and accelerate AI with new integrated hybrid cloud for edge capabilities that quickly help businesses grow and only pay for what they need.

Lenovo Unveils New Data Management Solutions to Enable AI Workloads

 Lenovo announced its next wave of data management innovation with new ThinkSystem DG Enterprise Storage Arrays and ThinkSystem DM3010H Enterprise Storage Arrays, designed to make it easier for organizations to enable AI workloads and unlock value from their data. Also announced are two new integrated and engineered ThinkAgile SXM Microsoft Azure Stack solutions, enabling a unified hybrid cloud solution for seamless data management. As businesses continue to scale their operations to address growing data, security and sustainability requirements, the new Lenovo flash solutions provide customers with an accelerated path to deploy AI workloads efficiently and with added security features from edge to cloud, enabling workload consolidation and mobilizing faster insights fortified with ransomware protection.

Lenovo Grows AI Infrastructure Revenue to Over US$2 Billion and Brings AI to the Data

• Lenovo simplifies deployment for businesses with additional US$1 billion investment in portfolio expansion and one-stop enablement wherever data resides, from the pocket to the cloud

• Lenovo AI Innovators program delivers 150+ turnkey solutions, helping businesses implement generative AI, immersive metaverse simulations and cognitive decisions at scale

• Expanded AI-ready portfolio of smart devices and edge-to-cloud infrastructure includes new Lenovo ThinkEdge and ThinkSystem platforms purpose-built for enabling AI workloads

Lenovo Launches Digital Workplace Solutions to Boost Employee Experience and Increase Productivity

Hybrid work styles preferred by employees create security and other workplace challenges that distract leaders from the core business 

• End-to-end Lenovo support, guidance and managed services create competitive advantages from hybrid workforce/workplace challenges

• Lenovo DWS helps companies improve operational cost, enhance productivity and improve security posture through end-to-end services built on a state-of-the-art platform and deep expertise in technology

At a time when companies are struggling to find customized technologies to address their unique business requirements, Lenovo Solutions and Services Group (SSG) is launching Digital Workplace Solutions (DWS), a new managed services portfolio of intelligent tools and systems. DWS delivers work-related technology, security, efficiency, and employee satisfaction so that executives can deliver on key performance outcomes.

Lenovo Unveils Next Generation of Intel-Based Smart Infrastructure Solutions to Accelerate IT Modernization

The post Top 10 News of Lenovo in 2023 appeared first on AiThority.

Vision-Based Speed Detection Algorithms

There are at least two primary reasons why it is becoming more and more crucial to precisely estimate the speed of road vehicles. To start, there has been a noticeable uptick in the number of speed cameras deployed around the globe in recent years. This is likely due to the widespread belief that enforcing reasonable speed limits is a great way to make roads safer for everyone. In addition, smart cities rely heavily on traffic monitoring and forecasting in road networks to improve traffic, pollution, and energy consumption.

One of the most important metrics for traffic conditions is vehicle speed. There are a lot of obstacles to overcome with vision-based systems when it comes to accurate vehicle speed detection, but there are also a lot of potential benefits, like a significant drop in costs (because range sensors aren’t needed) and the ability to correctly identify vehicles.

Video camera input data is the foundation of vision-based speed detection algorithms. There will be a photo album starting with the initial reveal and ending with the final reveal for every car. Factors such as vehicle speed, focus length, frame rate, and camera orientation relative to the road determine the total number of usable photographs.

Features:

• Variously known as traffic surveillance cameras or traffic CCTV, traffic cameras record traffic events live. Automatic or manual (visible inspection by a human operator) traffic flow, congestion, and accident monitoring is possible with their help. These cameras are often infrastructure- or drone-based and set up at a distance from the flow of traffic.
• Cams that record vehicles’ speeds are often called traffic enforcement cameras. In common parlance, they are a tool for keeping tabs on cars’ speeds apart from the actual speed sensor (be it radar, laser, or vision). The word “camera” comes from the fact that any system that uses radar or lasers to capture images of the vehicle also uses a camera. The word “speed camera” is used here in its most basic sense, meaning systems that measure speed using vision. Compared to the traffic cameras, their placement is typically more advantageous.

The post A Computer Vision System Accurately Computes Real-Time Vehicle Velocities appeared first on AiThority.

Cornell and Open University of the Netherlands scholars have developed a programmable network model to modernize the internet’s architecture through software-defined networks. This model allows researchers and network administrators to customize packet scheduling, the air traffic control mechanism built into network switches. The research team focused on the network switch, which powers networks and the internet, to design the next generation of networking hardware-software.

Small pizza box-sized switches govern network data flow and connect devices to a computer network. They schedule packets, which routes data through a network. The switch processes data from thousands of network users, including emails, news site visits, and Zoom calls. The switch’s packet scheduler prioritizes and arranges data clusters according on network manager policies. Finally, the switch forwards packets to nearby switches until they reach the end user’s device.

Researchers claimed changing this air traffic control procedure has been impossible because manufacturers encode scheduling specifications into the switch. Once placed in new network switches, the team’s methodology would allow network administrators to customize the switch’s packet-scheduling software, building on MIT and Stanford researchers’ 2016 solution.

The post Reimagining Packet Scheduling: Elevating Network Switches to Air Traffic Control Experts appeared first on AiThority.

One may make the case that creators, authors, and artists should have some say over who uses and profits from their work. This is typically accomplished via copyright laws. In most cases, the legal notion of “individual intellectual effort” is used by these statutes to establish authorship. In other words, the artist must have infused their work with sufficient originality and imagination to set it apart from previous works. But how can a person accomplish this? Some contend that, in contrast to AI, humans possess a unique quality that enables us to produce “new” works of art.

The IP battle between humans and AI

When it comes to intellectual property law, many jurisdictions have ruled that only “real humans” can be inventors, creators, or authors. However, when AI is involved, it’s not always apparent who is regarded as the author of a piece. Currently popular generative AI solutions take text suggestions as input and output what the user wants. Did a human put in enough labor to be deemed the author, inventor, or creator of the output work when they entered a specific set of prompts into an AI tool? In such case, where did the creative energy and originality originate from if the work is not plagiarized?

Many issues arise for those making and utilizing these technologies as a result of this line of thinking, particularly when trying to establish ownership. Generally speaking, it’s bad for the IP system as a whole. What, however, happens when an AI tool reaches the point where it is as knowledgeable as a human and has accumulated all the facts and experiences that a human could ever have? Similar to how a chess computer can anticipate every possible move a grandmaster would make, the AI would be capable of solving every difficulty that a person might think of. As a result, practically no new ideas are generated today, unless the human creator possesses exclusive, non-disclosable data.

Avoiding intellectual property problems with generative AI

You can take immediate, actionable actions to guarantee that anything created with the aid of generative AI will be credited to you as the creator, author, or inventor. The most critical thing is to keep track of when and how you employ AI technologies, as well as the data you use to obtain results. The newest generation of AI tools require you to document the prompts you use, together with the date and version of the tool, so they can be properly tracked. This might be very important later on when you need to prove that you were the rightful creator or inventor by demonstrating that enough “intellectual effort” was put into it.

It is important to ensure that you possess adequate rights to the datasets utilized for training new AI tools before beginning development. By doing so, you may rest assured that your tool’s underlying AI model will not mistakenly generate derivative works that violate the rights of others. The number of governments mandating the sharing of training datasets is expected to grow over time.

The post AI Revolutionizing Invention: Embracing the Ever-Evolving Realm of Innovation appeared first on AiThority.

Massive Artificial Intelligence Database LAION

Until recently, experts focusing on child abuse believed that unregulated AI tools could only create child abuse images by merging their knowledge of adult pornography and non-harmful child photos.

The massive artificial intelligence database LAION contains over 3,200 photos of children who may have been sexually abused. This database has been used to train top AI image-makers like Stable Diffusion. Together with the Canadian Centre for Child Protection and other anti-abuse organizations, the Stanford University watchdog group was able to detect the inappropriate content and notify the authorities of the original photo connections. It claimed that about 1,000 of the pictures it discovered had been independently verified.

Read: 4 Common Myths Related To Women In The Workplace

Malicious Content Creation

There was a lightning-fast reaction. While the Stanford Internet Observatory’s study was expected to be released, LAION informed The Associated Press that it would be temporarily withdrawing its datasets. In a statement, the nonprofit organization known as LAION (Large-scale Artificial Intelligence Open Network) stated that it “has a zero-tolerance policy for illegal content, and in an abundance of caution, we have taken down the LAION datasets to ensure they are safe before Republishing them.”Stability AI, a London-based firm that makes the Stable Diffusion text-to-image models, is a notable LAION user who contributed to the development of the dataset. While recent updates to Stable Diffusion have significantly reduced the likelihood of malicious content creation, a previous version from last year—which Stability AI denies releasing—is still embedded in various applications and tools and is reportedly “the most popular model for generating explicit imagery,” as stated in the Stanford report.

Even though it isn’t always obvious, the LAION database is a source for many text-to-image generators. The developers of DALL-E and ChatGPT, OpenAI, have said that they do not employ LAION and have adjusted their models to reject requests for sexually explicit material involving children.

Read: Top 10 News of AWS in 2023

The Stanford Internet Observatory Thoughts

The Stanford Internet Observatory has called for more extreme measures because it is difficult to clean up the data retroactively. “Delete them or work with intermediaries to clean the material” is one option for everyone who has constructed training sets using LAION-5B, which is named for the over 5 billion image-text pairs it contains. Another option is to completely remove an earlier version of Stable Diffusion from the internet, leaving only the most obscure places for it. For instance, Thiel criticized CivitAI, a platform popular among those who create AI-generated pornography, for what he perceived as a lack of safeguards that would prevent the creation of photos of minors. Hugging Face, an AI startup that provides model training data is also urged to improve its reporting and removal processes in the study.

In light of safeguards provided by the federal Children’s Online Privacy Protection Act, the Stanford paper raises the question of whether any images of children, no matter how innocent, should be inputted into AI systems without the permission of their families. To identify and remove child abuse content, tech corporations and child protection organizations already issue films and photographs a “hash”—unique digital signatures. If you believe Portnoff, you may apply the same idea to abused AI models.

Read:Top 10 News of Google in 2023

The post Unveiling the Disturbing Truth: AI Algorithms’ Training with Explicit Images of Young Ones appeared first on AiThority.

Thaler’s Case Reached the Highest Court

In a landmark decision, the highest court in the United Kingdom rejected the idea of artificial intelligence programs being recognized as patent inventors, thereby putting robots on par with humans. The Supreme Court of Britain denied the patents that Stephen Thaler, founder of Imagination Engines Inc., had requested, which would have identified his artificial intelligence computer DABUS as the creator. Judgment on Thaler’s appeal was unanimously rejected by the judges, who ruled that “DABUS is not a person at all,” by the requirements of patent laws.

AI Machine DABUS

The developer from the US asserts his entitlement to the inventions made by the artificial intelligence machine DABUS, which he says built an autonomous food or drink container and a light beacon. Since DABUS was not a natural person, the IPO determined in December 2019 that the specialist could not formally name it as the inventor on patent applications. Both the high court and the court of appeals upheld the decision in July 2020 and July 2021, respectively. The five-judge bench of the Supreme Court unanimously rejected Thaler’s argument following a March hearing.

The courts were not required to decide on whether the AI actually generated its inventions; the DABUS issue centered on how applications are made under the Patents Act 1977 legislation. Patents are legal protections for innovative and useful innovations that meet specific criteria set down by the government. These criteria include being technically sound, novel, and capable of being manufactured or used. AI advancements, like OpenAI’s ChatGPT technology, have recently been under investigation for a variety of reasons, including the possible effects on education, the dissemination of disinformation, and the future of employment. In this context, Thaler’s case reached the highest court.

The post The Enigma of AI Creations: Defying Recognition as Patent Inventors appeared first on AiThority.