Info World

We need to have a frank conversation about large language models (LLMs). At their core, LLMs are nothing more than sophisticated memorization machines, capable of reasonable-sounding statements, but unable to understand fundamental truth. Importantly and despite the fervent hopes of many, they are far from delivering or even prefiguring artificial general intelligence (AGI). The hype surrounding LLMs has reached stratospheric levels, fostering a misguided belief in their potential as AGI precursors.

We find ourselves at a critical juncture where the erroneous linkage between LLMs and AGI threatens to slow down—not accelerate—genuine progress in artificial intelligence. The clamor for LLMs to evolve into AGI solutions epitomizes tunnel vision at its finest. Consider the vast investments poured into training ever-larger models, yielding only marginal improvements in tasks that are not text-based. Let’s face it: LLMs are not learning how to do mathematics. Their forte lies in tackling statistical text tasks with finesse. It’s imperative that we recalibrate expectations and acknowledge that although LLMs excel in certain domains, they fall short in others.

To read this article in full, please click here

Just a few years ago, a prompt was something English teachers used for homework assignments, which filled up weekends and kept students inside on sunny days. Now it seems we’re all teachers, tasked with distributing perfect prompts that direct large language models to do our bidding. These prompts are also endowed with the power to ruin weekends, but it’s not the machines that are suffering.

The power of prompts can seem downright magical. We toss off a few words that approximate a human language and, voila! Back comes a nicely formatted, well-structured answer to whatever question we asked. No topic is too obscure and no fact is out of our reach. At least as long as it’s part of the training corpus and approved by the model’s shadowy controllers.

To read this article in full, please click here

There’s significant buzz and excitement around using AI copilots to reduce manual work, improving software developer productivity with code generators, and innovating with generative AI. The business opportunities are driving many development teams to build knowledge bases with vector databases and embed large language models (LLMs) into their applications.

To read this article in full, please click here

Many business processes involve communicating with external parties such as suppliers, logistics providers, insurance providers, and government institutions. Unlike internal communications, these inter-organizational communications usually have to follow various regulations and standards, and could introduce delays if one party doesn’t comply. Therefore, effective business-to-business (B2B) integrations are a key part of most digital transformation projects.

Let’s take the fictional example of an electronic equipment manufacturer named Smartware. Smartware supplies equipment for multiple retailers and depends on multiple suppliers and logistics partners for fulfilling orders. Their order processing workflow can be depicted as follows:

To read this article in full, please click here

The Eclipse Foundation announced that it is partnering with the Apache Software Foundation and other open source foundations to establish common specifications for secure software development based on existing open source best practices.

In an April 2 blog post, Eclipse said that the goal of the initiative was to meet the challenges of cybersecurity in the open source ecosystem and demonstrate cooperation with the European Union’s Cyber Resilience Act (CRA). Participants include Apache, Eclipse, the Rust Foundation, the PHP Foundation, the Blender Foundation, the OpenSSL Software Foundation, and the Python Software Foundation.

To read this article in full, please click here

Microsoft has updated its Azure AI Search service to increase storage capacity and vector index size at no additional cost, a move it said will make it more economical for enterprises to run generative AI-based applications.

Formerly known as Azure Cognitive Search, the Azure AI Search service connects external data stores containing un-indexed data with an application that sends queries or requests to a search index. It consists of three components—a query engine, indexes, and the indexing engine—and is mostly used in retrieving information to enhance the performance of generative AI, a process known as retrieval-augmented generation (RAG).

To read this article in full, please click here

Cohere has unveiled its latest large language model (LLM), Command R+, which is engineered to enhance enterprise workflows and applications. 

The company said the new model is its most advanced and scalable LLM yet. Building on the foundations of the earlier Command R model, Command R+ boosts performance for various enterprise tasks, including data categorization and workflow automation, the company said.

Although Cohere said in March that it would train and deploy its models on Oracle Cloud Infrastructure (OCI) under its partnership with Oracle, it has chosen to make Command R+ available first on Microsoft Azure. The software is set to launch on OCI shortly, with plans to expand to more cloud platforms in the upcoming weeks, said Cohere spokesman Kyle Lastovica. It’s already available through Cohere’s own hosted API.

To read this article in full, please click here

The latest data from the IDC Worldwide Quarterly Enterprise Infrastructure Tracker paints a compelling picture of growth in cloud infrastructure sales on demand. The fourth quarter of 2023 saw an 18.5% year-over-year increase in spending on compute and storage infrastructure for cloud deployments. It is a significant shift in the technological landscape, where AI is now front and center in the push to find cloud infrastructure to run it.

The spending surge indicates shifting budgets; a contrasting trend is the decline in the total number of units shipped. IDC says this shows a strategic move towards high-capacity, GPU-heavy servers with higher average price tags favored by hyperscalers. The idea is that these servers can do more than those driven by traditional CPU-based servers, thus, you need fewer.

To read this article in full, please click here

With the release of .NET 8, Microsoft introduced a plethora of new features and enhancements in the .NET Core and ASP.NET Core frameworks. One such feature is the SearchValues class, which marks a significant step forward in efficiently fetching data from data sets.

SearchValues is a new type introduced in .NET 8 designed to improve application performance. By using optimization techniques like vectorization and hardware acceleration, SearchValues delivers speed enhancements while seamlessly blending with .NET Core and ASP.NET Core.

In this article, we’ll explain how you can use SearchValues to improve the speed of searches in .NET Core applications.

To read this article in full, please click here

Microsoft has unveiled Visual Studio Code 1.88, also known as the March 2024 release of the company’s popular code editor. The update brings capabilities such as a test coverage API, custom editor labels, and locked scrolling.

Introduced on April 4, Visual Studio Code 1.88 can be downloaded for Windows, Linux, or Mac from the project website.

In this release, Microsoft has finalized its Test Coverage API, bringing native coverage support to VS Code. If a developer’s testing system supports it, coverage can be accessed using the new Run With Coverage button. Using test coverage requires that your VS Code extension implements the new API.

To read this article in full, please click here

JetBrains has added full-line code autocompletion to its family of IDEs (integrated development environments), powered by locally run AI models. Full line code complеtion is available for Java, Kotlin, Python, JavaScript, TypeScript, CSS, PHP, Go, and Ruby with the 2024.1 versions of the respective IDEs.

To read this article in full, please click here

Database and related services provider DataStax is acquiring Logspace, creator of the open source package Langflow, for an undisclosed sum, in order to help developers build generative AI applications faster, it said Thursday.

Langflow is an open-source, web-based no-code graphical user interface (GUI) that allows developers to visually prototype LangChain flows and iterate them to develop applications faster. LangChain is a modular framework for Python and JavaScript that simplifies the development of applications that are powered by generative AI language models or LLMs.  

To read this article in full, please click here

Once you get past the chatbot hype, it’s clear that generative AI is a useful tool, providing a way of navigating applications and services using natural language. By tying our large language models (LLMs) to specific data sources, we can avoid the risks that come with using nothing but training data.

While it is possible to fine-tune an LLM on specific data, that can be expensive and time-consuming, and it can also lock you into a specific time frame. If you want accurate, timely responses, you need to use retrieval-augmented generation (RAG) to work with your data.

The neural networks that power LLMs are, at heart, sophisticated vector search engines that extrapolate the paths of semantic vectors in an n-dimensional space, where the higher the dimensionality, the more complex the model. So, if you’re going to use RAG, you need to have a vector representation of your data that can both build prompts and seed the vectors used to generate output from an LLM. That’s why it’s one of the techniques that powers Microsoft’s various Copilots.

To read this article in full, please click here

At my company, Descope, we are building a customer authentication and identity management product, meaning the end user and the developer are two sides of the same coin. Providing a great developer experience—by enabling our customers to easily add auth flows and user management to their apps—leads to a great end-user experience as the customer’s customers seamlessly and securely log in.

This kind of virtuous cycle exists at many developer-focused companies. When building a successful developer-first business, it’s critical to tie together the similarities between the customer experience and the developer experience while clearly delineating the differences.

To read this article in full, please click here

Google has introduced Jpegli, a JPEG library for image encoding. The new library is intended to be faster, more visually pleasing, and more efficient than traditional JPEGs. Proponents of the technology said it has the potential to make the Internet faster and more beautiful.

Announced April 3 and accessible from GitHub, Jpegli maintains high backward compatibility while offering enhanced capabilities and a 35% compression ratio at high-quality compression settings, Google said. Jpegli works by using new techniques to reduce noise and improve image quality. New or improved features include adaptive quantization heuristics from the JPEG XL reference implementation, improved quantization matrix selection, calculation of intermediate results, and the possibility to use more advanced colorspace.

To read this article in full, please click here

GitHub Actions, an automated CI/CD platform for GitHub, has been enhanced for enterprise customers, with capabilities including stronger security and GPU-enhanced runners for machine learning.

GitHub announced updates to its hosted runner fleet for Actions on April 2.  To strengthen security, GitHub Actions now offers Azure private networking for GitHub-hosted runners. The feature combines compute-in-the-cloud with secure access and control over network security, eliminating the overhead of maintaining infrastructure. Hosted runners for every major operating system are intended to make it easy to build and test a project, which can be run directly on a virtual machine or a container.

To read this article in full, please click here

OpenTofu’s founders had a mission. Upset by HashiCorp licensing changes in August 2023 to its popular Terraform infrastructure-as-code tool, OpenTofu set out to be the “open source successor to the MPLv2-licensed Terraform,” further promising that it “will be community-driven, impartial, layered and modular, and backward-compatible.”

Hugely promising, but extraordinarily difficult to pull off. So difficult in fact, that OpenTofu may have illegally taken HashiCorp’s code to keep pace.

To read this article in full, please click here

Over the past decade, Rust has emerged as a language of choice for people who want to write fast, machine-native software that also has strong guarantees for memory safety.

Other languages, like C, may run fast and close to the metal, but they lack the language features to ensure program memory is allocated and disposed of properly. As noted recently by the White House Office of the National Cyber Director, these shortcomings enable software insecurities and exploits with costly real-world consequences. Languages like Rust, which put memory safety first, are getting more attention.

To read this article in full, please click here

Bun and HTMX are two of the most interesting things happening in software right now. Bun is an incredibly fast, all-in-one server-side JavaScript platform, and HTMX is an HTML extension used to create simple, powerful interfaces. In this article, we'll use these two great tools together to develop a full-stack application that uses MongoDB for data storage and Elysia as its HTTP server.

To read this article in full, please click here

Apache Spark is a data processing framework that can quickly perform processing tasks on very large data sets, and can also distribute data processing tasks across multiple computers, either on its own or in tandem with other distributed computing tools. These two qualities are key to the worlds of big data and machine learning, which require the marshalling of massive computing power to crunch through large data stores. Spark also takes some of the programming burdens of these tasks off the shoulders of developers with an easy-to-use API that abstracts away much of the grunt work of distributed computing and big data processing.

To read this article in full, please click here