Cloud Hosting

If 2022 was the year when AI broke through to become a society-changing technology, 2023 has been the year of AI breaking through in the enterprise. Put it this way: generative AI and large language models (LLMs) have become a familiar part of the lexicon for IT departments the world over. Your CIO is now more likely to namecheck ChatGPT than Kubernetes.

Already, its been a year of giant strides forward for AI in the enterprise, so here are five things weve learned so far.

In March, OpenAI announced an enterprise version of ChatGPT. But in this case, OpenAI was a fast follower not the first-to-market. Cohere, a Toronto-based company with close ties to Google, was already selling generative AI to businesses, as I discovered when I spoke to them that month.

Were working with developers in organizations, the AI/ML teams, to bring these capabilities into their organizations, Coheres CEO Martin Kon told us. He added that its approach is fundamentally different from OpenAIs. OpenAI wants you to bring your data to their models, exclusive to Azure. Cohere wants to bring our models to your data, in whatever environment you feel comfortable in.

So far, companies are mostly using generative AI to create semantic search engines for their own private data whether for internal use or for external customers. A related use case is knowledge management (KM). Imagine an employee being able to have a conversation with an AI that was trained on large language models based on company data.

One of the many new AI companies trying to achieve KM in chatbot form is Vectara. Its CEO Amr Awadallah told me that in five years, every single application whether that be on the consumer side or in the business/enterprise side will be re-architected in a way that is a lot more human in nature, in terms of how we express what we are trying to achieve and what were trying to do.

Last month, Google Cloud and Googles Workspace division announced a raft of new AI functionality. These included Generative AI App Builder, which allows organizations to build their own AI-powered chat interfaces and digital assistants, and new generative AI features in Google Workspace.

Google has coined a typically awkward term for applications powered by generative AI: gen apps. It claims that gen apps will become a third major internet application category, after web apps and mobile apps.

Im willing to wager that gen apps as a term will never take off but, regardless, Googles AI-powered tools will be well used within enterprise companies.

Likewise, Microsoft has been releasing new AI tools, such as Semantic Kernel (SK), described as an open-source project helping developers integrate cutting-edge AI models quickly and easily into their apps. SK is in many ways a classic Microsoft low-code tool, only it focuses on helping users do prompts for AI chatbots.

If you scan Stanfords HELM website, which measures LLMs in a variety of ways, youll see that these models vary greatly in size. There are, simply put, tradeoffs between model size and the speed it can work at.

OpenAI has several models, ranging in size from 1.3 billion parameters (its Babbage model) to 175B parameters (its DaVinci model).

Cohere goes as far as differentiating its model sizes as if they were Starbucks cups: small, medium, large, and extra large.

List of Coheres models in Stanford HELM directory

The primary models of OpenAI

However, Stanford also measures accuracy and in these statistics, size doesnt seem to matter much:

Stanford HELM tests for accuracy of ML models

In this new era of generative AI, frameworks like Ray are now just as important as Kubernetes in creating modern applications at scale. An open source platform called Ray offers a distributed machine-learning framework. Its being used by both OpenAI and Cohere to help train their models. Its also used in other highly-scaled products, such as Uber.

The company behind Ray is Anyscale, whose CEO Robert Nishihara told me earlier this year is very developer-centric. All of Rays functionality is designed to be easy to use for developers, he said. This differs from the Kubernetes user experience for devs, he noted, which is notoriously difficult. Ray was designed for Python developers, which is the main programming language being used in AI systems.

Anyscale co-founder Ion Stoica told me that Ray is like an extension of Python and like Python there are a set of Ray libraries that are targeted towards different use cases. The awkwardly named RLlib is for reinforcement learning, but there are similar libraries for training, serving, data pre-processing, and more.

Just as cloud computing ushered in a raft of big data solutions, generative AI is a catalyst for a new wave of data intelligence companies.

I recently spoke to Aaron Kalb, a co-founder of Alation, which styles itself as a data intelligence platform and promotes a concept it calls the data catalog. This combines machine learning with human curation to create a custom store of data for enterprise companies.

Kalb noted that both AI and the common enterprise acronym of BI (business intelligence) are garbage in, garbage out. He said that data intelligence is a layer that precedes AI and BI, that makes sure you can find, understand and trust the right data to put into your AI and BI.

In this context, he said, taking something like ChatGPT from the public internet and bringing it into the enterprise is very risky. He thinks that data needs to be, well, more intelligent before it is used by AI systems within an enterprise. Also, he doesnt think that the internet scale of ChatGPT and similar systems is needed in the enterprise. Every organization has its own terminology, he explained that could be industry terms, or things that are very specific to that company.

Its hard to believe that its been just a year since generative AI burst onto the scene. It all started with OpenAIs DALL-E 2, which was announced last April and launched as a private beta in July. DALL-E 2 is an image generation service powered by deep learning models, and it was a significant step forward for the industry. Also, last July, a company called Midjourney released its eponymous text-to-image generator.

The AI hype really ramped up in August, though, with the release of Stable Diffusion another deep-learning text-to-image generator. Unlike DALL-E 2 and Midjourney, Stable Diffusion had a permissive licensing structure. This was around the time people began paying more attention to the models behind all these services the LLMs. DALL-E 2 used a version of GPT-3, OpenAIs main LLM at the time.

But, in hindsight, we can see that the image generators were just the appetizer. At the end of November, OpenAI launched ChatGPT, a chatbot built on top of GPT-3.5. This was the catalyst for generative AI to enter the enterprise, and it has quickly taken over IT departments since then.

View post:
5 Key Learnings about AI and ChatGPT in the Enterprise - The New Stack

There is a lot of buzz around artificial intelligence and its different algorithms. We all are quite aware that machines along with specific computer algorithms can do wonders in our homes, offices or at workplaces. With the advancement of technology, one must know the main reasons behind several hi-tech inventions and innovations, is the new concept of deep learning.

Ever wondered how Netflix or Amazon showing recommendations based on your preferences, or how Siri is able to respond to each and every request. It means, deep learning algorithms are at work.

Right now, it may leave you a little perplexed, but heres a simple guide on deep learning, how it works and how it is deeply associated with the AI world. Forbes Advisor India has broken down this particular branch of AI in most simple terms.

Deep learning runs many artificial intelligence (AI) applications and services. It helps in adding intelligence and improving automation to the existing AI enabled products. DL is that part of AI which helps in performing analytical and physical tasks without any sort of human intervention.

In short, deep learning is a complex technique of machine learning, which instructs computers to learn or respond as to what naturally comes to humans. So, whether it is driverless cars, hands-free speakers, voice recognition in phones, tablets, TV or watches, deep learning is a major force behind all these breakthrough innovations.

In the concept of deep learning, the computer learns to perform on the basis of direct data feed such as image, text or sound. Such models are capable of achieving super accurate results and sometimes much better and more efficiently than human beings. Models based on deep learning uses a large set of data which requires high computation power and responds accurately via using a neural network which contains multiple layers like that of the humans brain.

In nutshell, deep learning sits inside of machine learning, which sits inside of artificial intelligence.

Artificial Intelligence: The development of a computer system which is able to perform all the given tasks at par with human intelligence.

Machine Learning: It is a subset of AI which contains statistical algorithms which enable machines to improve the tasks with experience. As opposed to deep learning, machine learning models need human intervention to improve accuracy.

Deep Learning: It is a kind of machine learning which has a human brain-like structure. It works on the basis of logical assembly of algorithms which are known as neural networks.

In the world of efficiency and accuracy, deep learning has made a notable and dominant position than ever before. Deep learning applications are used in various industries from healthcare, automated driving, medical devices, aerospace and defense, electronics and industrial automation.

Deep learning is extensively used in automated hearing, speech recognition, language translation, digital assistance, etc.

To give such accurate results, DL requires a large amount of labeled data and high computation power. High-performance GPUs have a perfect and ideal architecture which has been proved efficient for deep learning to perform. When combined with clusters or cloud computing, this technology enables teams to reduce training time for a deep learning network from weeks to hours or may be lesser than this.

For example, driverless car development requires billions of images and hours and hours of video, which help deep learning to automatically detect objects such as pedestrians, stop signs and traffic lights.

Mostly, all the deep learning-based models use architectures related to neural networks. The term deep is associated with this technique as it refers to the multiple number of hidden layers which are present in a neural network.

Deep learning models have been given proper training by using a large subset of labeled data and neural network architectures, which in turn helps these models to learn directly from that data without the need of any human intervention.

Neural networks are generally organized in multiple layers consisting of a different set of interconnected nodes. It is to be noted that these networks have the ability to have tens or hundreds of hidden layers.

Most deep learning features use the transfer learning approach, a procedure which involves fine-tuning a pretrained model. However, the relevant features are not pre trained as they are learned while the whole network trains on a collection of images. This feature includes automated extraction which makes deep learning models very accurate.

The key advantage of deep learning models is that they continue to improve as the size of the data upsurges. The more practice deep-learning algorithms get, the better they become.

Deep learning algorithms are very complex in their nature. There are different types of neural networks which helps in addressing the specific problems or datasets, such as:

Convolutional Neural Network (CNN): This kind of neural network has a certain degree of complexity which is seen in human brains. Neural network is not just made up of one layer but it consists of various layers which are also known as additional convolutional layers or pooling layers. It is to be noted that each layer plays a very crucial role in grasping the data and thus reaching a final conclusion which means identifying bigger portions of the image.

The first layer focuses on simple and basic features, such as colors and as the data progresses the layers of the network start to recognize much bigger elements of the object until it finally identifies the final object.

Recurrent Neural Network (RNN): Recurrent neural networks use consecutive data or time series data to resolve common problems seen in language translation and speech recognition which are used in various applications such as Siri, voice search, and google translate.

Similar to convolutional neural networks (CNNs), the RNN uses training data to learn. They distinguished the fed data by their memory as they take cues from prior inputs, which in turn influences the current input and output.

As we have already learnt, deep learning requires a massive amount of computation power. For this, high performance graphical processing units or (GPUs) are considered as ideal as they are able to handle and process quite a large volume of calculations and patterns with abundant memory available. It is to be noted that managing multiple such GPUs need high requirement of of internal resources which can be incredibly costly

Deep learning plays an important role in AI predictive modeling. It collects massive amounts of data and analyzes it to create multiple predictive models by understanding various patterns and trends within the data. Deep learning models make it very fast and easy to construct large amounts of data and form them into meaningful information. It is widely used in multiple industries, including automatic driving and medical devices. It is just a matter of time as this technology continues to mature.

Read more:
What Is Deep Learning AI & How Does It Work Forbes Advisor INDIA - Forbes

With quantum machine learning, the internet of things can become even more powerful, enabling people to create more efficient and safer systems.

FREMONT, CA: The Internet of Things (IoT) is altering how people interact with their surrounding environment. From intelligent homes to autonomous vehicles, the possibilities are limitless. Researchers are investigating the possibility of merging IoT with quantum machine learning (QML) to create even more powerful and efficient systems.

QML is an artificial intelligence (AI) that processes data using quantum computing. It offers the ability to provide quicker and more precise decision-making than conventional AI. Researchers hope to create a potent new data analysis and prediction tool by merging it with the IoT.

QML and IoT could be combined to create smarter, more efficient systems for various applications. For instance, it might optimize city traffic flow by forecasting traffic patterns and modifying traffic light timing accordingly. It could also be utilized to optimize building energy consumption and monitor and predict disease spread

IoT facilitates the huge potential of QML enabled by IoT. It could transform how people interact with the environment around them and create new opportunities for data analysis and forecasting. As researchers continue to investigate the possibilities, it is evident that this technology can alter the way of life.

Using the IoT to Advance QML

The IoT is altering how people interact with their surrounding environment. IoT technology's potential applications appear limitless, from intelligent homes to self-driving vehicles. Now, scientists are investigating how IoT can transform QML.

QML is a fast-developing research topic that blends quantum computing capabilities with machine learning methods. QML can enable robots to learn more effectively and precisely than ever before by harnessing the potential of quantum computing.

The IoT is ideally suited to supporting QML applications. IoT devices can collect and communicate vast quantities of data, which can be utilized to train and optimize machine learning algorithms. In addition, IoT devices can be used to monitor and control the environment in which QML algorithms are deployed, ensuring that they operate under optimal conditions.

Also, researchers are investigating how IoT devices might be leveraged to enhance the security of QML applications. IoT devices can identify and prevent harmful attacks on QML systems by harnessing the power of distributed networks. IoT devices can also be used to monitor the performance of QML algorithms, enabling the immediate identification and resolution of any problems.

The potential uses of the IoT for QML are vast, and researchers are just beginning to investigate them. By leveraging the power of the IoT, researchers are paving the way for a new era of QML that might transform how people interact with the world.

Originally posted here:
Exploring the Possibilities of IoT-Enabled Quantum Machine Learning - CIOReview

Theres no question that machine learning operations (MLOps) is a burgeoning sector. The market is projected to reach $700 million by 2025 almost four times what it was in 2020.

Still, while technically sound and powerful, these solutions havent generated the expected revenue, which has raised concerns about future growth.

I can understand the pessimism surrounding the space, as I spent the first 20 years of my career effectively building internal MLOps tools at an esteemed investment management firm. More recently, Ive invested in MLOps startups, but they have been slow to achieve the level of revenue that I would have expected. Based on both my positive and negative experiences with MLOps, I understand why these startups have struggled and why they are now poised for growth.

MLOps tools are critical to companies deploying data-driven models and algorithms. If you develop software, you need tools that allow you to diagnose and anticipate problems with software that could cause you to lose meaningful revenue due to its failure. The same is true for companies that build data-driven solutions. If you dont have adequate MLOps tools for evaluating models, monitoring data, tracking drift in model parameters and performance, and tracking the predicted vs. actual performance of models, then you probably shouldnt be using models in production-critical tasks.

However, companies deploying ML-driven solutions without deep knowledge and experience dont recognize the need for the more sophisticated tools and dont understand the value of the low-level technical integration. They are more comfortable with tools operating on externalities, even if they are less effective, since they are less intrusive and represent a lower adoption cost and risk if the tools dont work out.

On the contrary, companies with ML teams who possess deeper knowledge and experience believe they can build these tools in-house and dont want to adopt third-party solutions. Additionally, the problems that result from MLOps tools shortcomings arent always easy to identify or diagnoseappearing as modeling versus operations failures. The outcome is that companies deploying ML-based solutions, whether technically sophisticated or inexperienced, have been slow to adopt.

But things are starting to change. Companies are now recognizing the value of sophisticated, deeply integrated MLOps tools. Either they have experienced problems resulting from not having these tools or they have seen competitors suffering from their absence in many high-profile failures, and are now being forced to learn about the more complex MLOps solutions.

Those MLOps companies that have survived the revenue winter so far should see a thawing of the market and a growth in sales opportunities.

Companies selling superficial solutions will start losing business to more integrated solutions that are harder to understand and adopt, but provide more monitoring, debugging, and remediation services for their customers. MLOps software developers should keep the faith that building powerful software that solves problems in a deeper and more thorough way will win out in the long run over simple solutions that give immediate payoffs but dont solve the full breadth of problems their customers are facing.

Read more:
Why Sophistication Will Win Out In The Machine Learning Ops Sector - Unite.AI

The field of technology law has seen significant growth over the past few years in South Africa. With the rapid pace of technological advancements and the increasing reliance on technology in various industries, the legal system must keep up with these changes. In this article, I will explore the future of tech law in South Africa.

One of the areas where technology law is likely to see development in South Africa is the regulation of data privacy. The Protection of Personal Information Act (PoPIA) protects personal information and regulates the processing of personal data. However, with the rise of big data and the increasing use of technology in various industries, the legal framework surrounding data privacy will likely evolve in the coming years. This may include changes to PoPIA itself, as well as new legislation and case law that addresses emerging issues in data protection. These issues include, but are not limited to

Another area where tech law will likely see development is regulating artificial intelligence (AI) and machine learning. As AI and machine learning become more widespread, concerns exist about their potential impact on the following areas:

In South Africa, no specific legislation currently governs the use of AI and machine learning. Therefore, to address these concerns, it is important for South African organisations and policymakers to prioritise privacy, security, and ethical considerations when developing and implementing AI and machine learning systems. This may involve developing robust data protection policies, ensuring adequate cybersecurity measures are in place, and promoting transparency and accountability in AI and machine learning systems.

The role of ChatGPT in the future of legal research and analysis cannot be overstated. As a large language model, ChatGPT has the potential to revolutionise the way legal research is conducted in South Africa. With the increasing volume of case law and legislation, it can be challenging for legal practitioners to keep up with developments in the field. ChatGPT can quickly and accurately analyse large volumes of legal texts, allowing legal practitioners to identify relevant case law and legislation. Additionally, ChatGPT can assist in legal writing by providing suggestions for legal arguments based on its analysis of previous case law and legislation.

In conclusion, the field of technology law in South Africa is likely to see significant growth and development in the coming years. With the rapid pace of technological advancements and the increasing reliance on technology in various industries, the legal system must keep up with these changes. The regulation of data privacy and AI and machine learning are two areas where tech law is likely to see development. Case law and legislation will be crucial in shaping these technologies' legal frameworks. ChatGPT also has the potential to revolutionise legal research and analysis in South Africa and will undoubtedly play a significant role in the future of tech law in the country.

Read the original here:
How AI, machine learning and ChatGPT are changing the legal system - Bizcommunity.com