How To Accelerate Software Development With Generative AI

How To Accelerate Software Development With Generative AI

By Remo Peduzzi

In the span of just a few years, generative AI has transformed how organizations build products, create content and resolve problems. The majority of business and technology leaders using GenAI are focusing on efficiency and cost-effectiveness gains, according to a 2024 Deloitte survey, with 91% of respondents reporting that they “expect generative AI to improve their organization’s productivity.”

For software developers, generative AI is a game changer, though I think the term “artificial intelligence” can be misleading. It’s not actually intelligent; it crunches numbers, recognizes patterns and processes data at a speed beyond human capabilities. It’s no replacement for human expertise and execution, but it can take the lead on mundane, repetitive tasks, allowing teams to dedicate more time and energy toward critical thinking, problem-solving and collaboration.

A 2023 McKinsey study showed that generative AI tools can significantly improve software developers’ productivity on common tasks: “Documenting code functionality for maintainability (which considers how easily code can be improved) can be completed in half the time, writing new code in nearly half the time, and optimizing existing code (called code refactoring) in nearly two-thirds the time.” And GitHub’s research found that Copilot, its popular AI development tool, helps developers complete tasks up to 55% more quickly.

By using GenAI tools on my own development team, I’ve discovered that they are most valuable in two key applications.

• Code suggestion and autocompletion: AI can analyze developers’ code as they work, automatically generating recommendations for code snippets or complete functions based on context and input. Autocompletion reduces manual coding time, especially when there are blocks of repetitive code in a project.

• Code analysis and bug detection: Generative AI can quickly review code to detect errors or bugs early in the development process. AI tools can also learn from historical data and automate code pattern analysis, proposing modifications to improve code blocks.

Strategies For Implementing AI In Software Development

To maximize the benefits of AI in software development, I recommend the following four strategies.

1. Test And Evaluate Different Tools

Experiment with a variety of AI plugins and platforms to find one that best fits your team’s needs. Test each tool on basic tasks, then scale up to bigger projects, such as detailed code creation or bug analysis, to gauge where it offers the maximum benefit.

2. Create Better Prompts

The quality of your results depends on the quality of your prompts. Learn how and what to ask GenAI to get the most out of your programming tools. If you don’t get the answer you’re looking for with your first query, iterate to refine your prompt. Provide additional context and specific requirements to guide AI toward your desired outcomes.

3. Review Code Carefully

GenAI can greatly improve your efficiency in code creation, but it still needs human oversight. I see AI as just another tool in an experienced developer’s toolbox. You can’t expect a junior developer to write perfect code with GenAI, in the same way you can’t expect a junior marketer to write a flawless article using ChatGPT. You may end up with a solid structure and some sections that work well, but you’ll also likely have blocks that need to be debugged or rewritten. Have a senior developer on your team double-check everything AI creates to prevent errors.

4. Protect Sensitive Data

Be mindful of data privacy and security considerations when using GenAI for software development. Set clear team policies that prohibit sharing sensitive client data or proprietary information with AI providers that make queries publicly available.

While GenAI can significantly enhance your team’s productivity, it’s not a panacea. It’s a valuable tool when trained and used correctly. You can’t expect it to work miracles, but you can leverage it to save time and money throughout the development pipeline. Set realistic expectations, and use AI tools strategically and thoughtfully—in conjunction with human expertise and oversight—to deliver software solutions more efficiently than ever before.

Read the full article

How AI can transform a burdensome and complex manufacturing environment

How AI can transform a burdensome and complex manufacturing environment

By Grace Nam

The manufacturing sector, known for its intricate processes and high demands for precision, is on the brink of a revolution driven by, you guessed it, AI. This transformation represents more than just a technological upgrade; it signifies a comprehensive overhaul that aims to rectify long-standing inefficiencies, streamline operations, and significantly boost productivity.

AI’s integration into manufacturing processes enables real-time data analysis, predictive maintenance, and automation of repetitive tasks, leading to reduced operational costs and minimized human error.

Every day, professionals devote a significant portion of their time to handling documents and upkeeping tasks, such as finding the location of paper, ensuring the final version of documents, saving the documents/structuring the documents, and reviewing and correcting data.

As manufacturers adopt AI, they can expect enhanced efficiency and output and the ability to make more informed decisions and adapt swiftly to market changes, thereby securing a competitive edge in an increasingly dynamic industry.

The current state of manufacturing

recent study by SME, involving over 300 manufacturing professionals, highlighted a critical issue: Frequent work delays across various operational processes. About one-third of the respondents reported experiencing these delays several times a week. Such interruptions not only waste valuable time but also impact productivity and profitability. The study underscores the pressing need for solutions to alleviate these operational bottlenecks.

AI’s potential in streamlining operations

AI has the potential to revolutionize essential manufacturing functions, from sales and supply chain management to quality control and inventory management. By integrating AI, manufacturers can predict potential disruptions, optimize resource allocation, and ensure timely deliveries.

For example, AI algorithms can analyze both historical and current data to forecast demand accurately, thereby aligning inventory levels with market needs and reducing the risk of overproduction or stockouts. This capability represents a critical revolution for manufacturing, as it enables decisions to be based not just on outdated data but on real-time analysis of both past and present information.

Addressing workforce challenges

The manufacturing industry is also grappling with a shortage of STEM professionals and a lack of standardized processes. However, AI technology offers a promising solution by automating complex tasks that traditionally do not require specialized skills to perform such as managing inventory data, job order, inspection, and customer service documentation that require manual administrative responsibilities.

AI can streamline rule-based processes, relieving process experts and employees of repetitive administrative tasks and allowing them to focus on more strategic and value-added activities to perform areas that require technical knowledge. By leveraging AI to automate these tasks, manufacturers can address the shortage of skilled professionals and also enhance the capabilities of their existing workforce.

While there are concerns about AI leading to job displacement, the reality is that AI will augment the human workforce. By automating traditional processes in manufacturing, AI frees employees to engage in higher-level activities that require creativity and problem-solving that involve more of human nature and expertise.

This shift enhances job satisfaction and drives innovation within the organization. As AI becomes more integrated into manufacturing, the demand for workers skilled in AI implementation and human-machine collaboration will increase, necessitating upskilling and reskilling initiatives, to the benefit of the workers who go through these processes. Ultimately, creating a better ecosystem with operational resilience and efficiency.

Enhancing interoperability

One significant hurdle in modern manufacturing is interoperability among different software systems. AI can solve this by enabling seamless communication between disparate platforms and teams. Through machine learning algorithms, AI can facilitate the exchange of information across systems, creating a cohesive and integrated operational environment. This interoperability is crucial for real-time decision-making and efficient process management.

Improving data accuracy and reducing errors

AI-capture tools are at the forefront of transforming how manufacturers handle data. These tools can streamline vast amounts of content, extract relevant information, and automatically organize it. This capability is particularly valuable in managing compliance documentation, which is often complex and time-consuming. By automating data capture and classification, AI ensures that all documents are easily searchable and retrievable, significantly reducing the time spent on administrative tasks.

Manual data entry and processing are prone to human errors, which can lead to costly and time-consuming mistakes that can cause delays across an organization. With its advanced recognition capabilities, AI can accurately extract data from documents, minimizing the risk of errors. This improves data accuracy and enhances the reliability of the information used for decision-making. AI can automatically classify metadata, making managing and retrieving documents easier, thereby streamlining records management.

Scaling operations seamlessly

AI offers unparalleled scalability, allowing manufacturers to expand their operations without a corresponding increase in complexity. AI transforms records management, data/information governance to ensure organization positions themselves to protect their critical assets, and process automation by connecting various enterprise applications. This seamless scalability ensures that as manufacturing operations grow, AI systems can efficiently handle the increased data volumes and operational demands.

Improving compliance management

Compliance with industry regulations is a significant concern for manufacturers. AI simplifies compliance management by automating data capture and document management. AI-powered document management systems streamline the organization, retrieval, and updating of compliance-related documents, minimizing errors and facilitating timely audits. By reducing the burden of endless numbers of compliance requirements, AI allows manufacturers to focus on core operations and strategic initiatives.

Real-world benefits of AI in manufacturing

Implementing AI in manufacturing offers tangible benefits such as reduced operational costs, improved productivity, and enhanced data management. By automating tedious tasks, AI lowers labor costs and minimizes human error. It also eliminates bottlenecks and delays, fostering a more productive workforce by automating daily step-by-step tasks whether they are rule-based or decision-driven. Additionally, real-time data capture and analysis provide valuable insights for better decision-making and future planning.

AI promises to transform the manufacturing sector by addressing existing challenges and unlocking new opportunities for efficiency and growth. As the recent study by SME illuminated, approximately one-third of manufacturing professionals are experiencing delays several times a week.

Leveraging AI to mitigate these delays and optimize operational processes can significantly enhance productivity and reduce costs.

This not only streamlines operations but also increases contributions toward organizational savings and drives higher revenue, whether through intentional revenue growth strategies or simply by operating more efficiently.

By embracing AI, manufacturers can navigate the complexities of their environment and pave the way for a more innovative and resilient future.

Read the full article

AI skills can help you land a job or promotion faster—especially for Gen Z, says new research

AI skills can help you land a job or promotion faster—especially for Gen Z, says new research

By Morgan Smith

Artificial intelligence might replace some jobs but it can also give you a competitive edge in the workplace.

Nearly all executives — 96% — feel an urgency to incorporate AI into their business operations, according to a March 2024 Slack Workforce Lab survey of more than 10,000 professionals.

But if you feel conflicted about this new technology, you’re not alone. Researchers can’t seem to agree on if, and how, workers are using AI: Some reports claim that professionals are excited about and experimenting with AI, while others say most adults have not tried using AI tools on the job or do not trust them.

Regardless of where you stand on AI, people who don’t learn to use it risk losing career opportunities to those who do, new research from Microsoft and LinkedIn found.

AI skills could rival job experience in hiring decisions — and not just in tech

Close to 70% of leaders say they won’t hire someone without AI skills and would rather hire a less experienced candidate with AI skills than a more experienced person without them, according to the report, which surveyed more than 30,000 people in 31 countries.

“Learning basic AI skills — such as prompt engineering, machine learning or data literacy — is the best insurance to shortcut your competitiveness against people who might have more experience,” Aneesh Raman, a vice president and workforce expert at LinkedIn, tells CNBC Make It.

Some companies including Google and Amazon have announced investments in teaching their workforce AI skills, but such initiatives aren’t the norm: Only 25% of companies are planning to offer training on generative AI tools like ChatGPT and Microsoft Copilot, Microsoft and LinkedIn found.

There are dozens of free online courses people can use to learn AI skills offered by companies like IBM and Google and Ivy League institutions like Harvard University and the University of Pennsylvania.

The hype around AI is far from peaking — it’s just starting to build, according to Colette Stallbaumer, general manager of Microsoft Copilot and co-founder of Microsoft WorkLab.

Of course, Microsoft is betting big on AI. In May, the tech giant announced it will invest $3.3 billion over the next four years to build new cloud and AI infrastructure.

“Less than two years after generative AI burst onto the scene, we’re seeing this technology being woven into the fabric of work across a wide range of industries,” Stallbaumer says. “This is happening at a pivotal time where the pressure, volume and pace of work from the Covid-19 pandemic has hardly let up. Employees are overwhelmed and turning to AI for help.”

Generative AI tools in particular have seen a surge in workplace adoption, with usage doubling in the last six months, Microsoft and LinkedIn report.

It’s not just programmers and engineers experimenting with these tools: Architects, project managers and administrative assistants are among the professionals looking to build their AI aptitude the most.

Non-tech industries including health care, finance and marketing are adopting AI technologies at a rapid clip to streamline business operations and boost productivity, Stallbaumer adds, creating high demand and new job opportunities for professionals skilled in these tools.

Gen Zers could use AI to accelerate their careers

As more leaders demand AI skills in new hires, younger applicants with AI acumen stand to have greater access to job opportunities over their more experienced peers without those skills and accelerate their ascent up the corporate ladder. 

Gen Z employees, being digital natives, are more likely to use these tools at work than their millennial, Gen X and Baby Boomer colleagues, Microsoft and LinkedIn found.

What’s more, 77% of leaders say that early-career talent with AI skills will be given greater responsibilities at work, the Microsoft and LinkedIn data shows.

Raman says AI could also help young professionals move their careers forward by providing faster access to tailored career advice, market research and other data-driven insights that help them feel more confident and competent in their jobs.

Lydia Logan, IBM’s vice president of global education and workforce development, expects that the rapid integration of AI in the workplace will trigger significant changes to entry-level job responsibilities.

“When I think about the first job I had, a lot of what I was doing was answering the phone, organizing files, and that’s still the case for a lot of people,” she says. “Many of those administrative tasks that can now be automated with AI, which leaves room for entry-level workers to take on the kind of responsibilities someone one or even two levels above them on the corporate ladder might have.”

Read the full article

Move Over, ChatGPT: Here Are 5 AI Education Tools Loved By Teachers

Move Over, ChatGPT: Here Are 5 AI Education Tools Loved By Teachers

By Dan Fitzpatrick

Artificial Intelligence is inspiring educators all over the world.

I reached out to teachers on LinkedIn, X, and Facebook to find out which AI tools are having the most impact in classrooms. The overwhelming response is testament to how much educators are loving their new AI workmates. After exploring the first round of popular tools, let’s take a look at five more.

While many educators are using leading AI chatbots such as ChatGPT and Google Gemini, platforms designed specifically for educators and students are offering more specialized functionality.

Here are five more AI tools making waves in classrooms worldwide:

  • Magic School
  • Poe
  • Chat For Schools
  • Magic Media in Canva
  • Udio

With insights from educators who are leveraging their potential, let’s explore them in more detail.

Five AI Tools For Educators

MagicSchool

MagicSchool is a generative AI platform designed to assist educators with various tasks such as lesson planning, writing assessments and creating individualized education plans. It offers over 60 tools to streamline various processes and aims to save teachers significant amounts of time each week. The platform also offers a suite of tools specifically designed for students to enhance their AI literacy and learning experience.

Heather Brown, a K-5 math interventionist and STEAM teacher in Illinois, shared her enthusiasm: “I love that MagicSchool has so much of the prompting built in to help students truly engage with AI. The variety of ways it can be used is also incredible, from rap battles to research assistants to math review and beyond! The guidance it gives to students before proceeding into using AI is also a great starting point for teachers to talk about AI in a factual, unbiased way.”

I use MagicSchool’s tools when leading sessions with students to help them create AI bots for their own personal use. I’m constantly amazed by how they innovate, developing bots for themselves to help with studying, career guidance and even sleep management.

Poe

Poe is a versatile AI tool that allows students to create personal chatbots and explore different AI models. It allows customization of chatbot behavior and responses, providing a more tailored and interactive experience.

Jason Gulya, a professor of English at Berkeley College uses Poe in his classroom: “I currently teach my students to use Poe. It helps them to create their own personal chatbots. When they use it, they start to realize that AI isn’t magic, but a technology that allows them to build useful solutions around their own problems and interests.”

Chat For Schools

I received a lot of replies mentioning this tool. Chat for Schools by Skill Struck is tailored for K-12 classrooms, allowing students to engage with AI chatbots. Teachers can create custom tutors, monitor chat history and control the AI’s usage to prevent cheating and ensure appropriate interactions. The platform also integrates features like sentiment analysis and reading level adjustments to enhance the educational experience.

Devan Miller, a career and technical education teacher in Florida, praised Chat For Schools: “It allows me to create tutors for specific aspects of content that I would like my students to practice and learn while allowing me to monitor my students’ queries. The management system also flags anything that could prove to be inappropriate for school. It’s an amazing resource that I recommend be used by teachers!”

Magic Media in Canva

Magic Media, an app in the Canva platform, enables users to create images, graphics and videos from simple text prompts. Teachers can generate high-quality visual content without requiring extensive design skills. Users can further enhance their creations using Canva’s editing tools, which offer options for adding animations, transitions and other effects to make the final product more dynamic and engaging​.

Ainsley Messina, a technology integrator in New York City, shared her positive experience: “I love using Canva Magic Media tools with my 4th graders! These tools are fantastic for teaching students how AI can enhance photo editing and presentation creation. By using Magic Media, my students get an engaging introduction to the capabilities of AI, learning firsthand how it can be integrated into creative projects. It’s also an excellent way to highlight the importance of detailed prompt writing, helping them understand how precise instructions can yield the best results. Overall, it’s a fun and educational way to prepare them for the future of digital creativity.”

Magic Media is just one of many AI tools in Canva. Educators can access these with a free premium account.

Udio

Udio is a new AI platform designed to generate custom music tracks based on user preferences. You simply type in what kind of song you would like, select the genre of music and wait around 1 minute for a unique 30 second song. Advanced features allow users to create full length songs, include their own lyrics and even generate a song from an audio input.

Pravin Kaipa, an education specialist at an elementary school in California, uses Udio with his special needs students: “My students and I love using Udio because we generate songs to help us memorize concepts or understand new ones, and we have created everything from mnemonics to remember the prime numbers under 10 to the differences between potential and kinetic energy. They also loved using it to create positive self-image theme songs that I play when they finish a big presentation.”

Important Considerations For Educators

These five tools are just a few of the many platforms shared by educators around the world. To search for other AI platforms being used in education, many educators use the AI Educator Tools repository.

Before integrating any digital platform into your classroom, it is crucial to follow your organization’s procedures regarding data protection. Always seek guidance from the people responsible for this in your school, college, or university.

The teachers I spoke to are saving many hours per week using tools such as the ones above. When integrated in a safe way, AI has the potential to transform the practices of any teacher.

This could be you.

Read the full article

A technique for more effective multipurpose robots

A technique for more effective multipurpose robots

By Adam Zewe

With generative AI models, researchers combined robotics data from different sources to help robots learn better.

Four photos show, on top level, a simulation of a robot hand using a spatula, knife, hammer and wrench. The second row shows a real robot hand performing the tasks, and the bottom row shows a human hand performing the tasks.

Let’s say you want to train a robot so it understands how to use tools and can then quickly learn to make repairs around your house with a hammer, wrench, and screwdriver. To do that, you would need an enormous amount of data demonstrating tool use.

Existing robotic datasets vary widely in modality — some include color images while others are composed of tactile imprints, for instance. Data could also be collected in different domains, like simulation or human demos. And each dataset may capture a unique task and environment.

It is difficult to efficiently incorporate data from so many sources in one machine-learning model, so many methods use just one type of data to train a robot. But robots trained this way, with a relatively small amount of task-specific data, are often unable to perform new tasks in unfamiliar environments.

In an effort to train better multipurpose robots, MIT researchers developed a technique to combine multiple sources of data across domains, modalities, and tasks using a type of generative AI known as diffusion models.

They train a separate diffusion model to learn a strategy, or policy, for completing one task using one specific dataset. Then they combine the policies learned by the diffusion models into a general policy that enables a robot to perform multiple tasks in various settings.

In simulations and real-world experiments, this training approach enabled a robot to perform multiple tool-use tasks and adapt to new tasks it did not see during training. The method, known as Policy Composition (PoCo), led to a 20 percent improvement in task performance when compared to baseline techniques.

“Addressing heterogeneity in robotic datasets is like a chicken-egg problem. If we want to use a lot of data to train general robot policies, then we first need deployable robots to get all this data. I think that leveraging all the heterogeneous data available, similar to what researchers have done with ChatGPT, is an important step for the robotics field,” says Lirui Wang, an electrical engineering and computer science (EECS) graduate student and lead author of a paper on PoCo.

Wang’s coauthors include Jialiang Zhao, a mechanical engineering graduate student; Yilun Du, an EECS graduate student; Edward Adelson, the John and Dorothy Wilson Professor of Vision Science in the Department of Brain and Cognitive Sciences and a member of the Computer Science and Artificial Intelligence Laboratory (CSAIL); and senior author Russ Tedrake, the Toyota Professor of EECS, Aeronautics and Astronautics, and Mechanical Engineering, and a member of CSAIL. The research will be presented at the Robotics: Science and Systems Conference.

Combining disparate datasets

A robotic policy is a machine-learning model that takes inputs and uses them to perform an action. One way to think about a policy is as a strategy. In the case of a robotic arm, that strategy might be a trajectory, or a series of poses that move the arm so it picks up a hammer and uses it to pound a nail.

Datasets used to learn robotic policies are typically small and focused on one particular task and environment, like packing items into boxes in a warehouse.

“Every single robotic warehouse is generating terabytes of data, but it only belongs to that specific robot installation working on those packages. It is not ideal if you want to use all of these data to train a general machine,” Wang says.

The MIT researchers developed a technique that can take a series of smaller datasets, like those gathered from many robotic warehouses, learn separate policies from each one, and combine the policies in a way that enables a robot to generalize to many tasks.

They represent each policy using a type of generative AI model known as a diffusion model. Diffusion models, often used for image generation, learn to create new data samples that resemble samples in a training dataset by iteratively refining their output.

But rather than teaching a diffusion model to generate images, the researchers teach it to generate a trajectory for a robot. They do this by adding noise to the trajectories in a training dataset. The diffusion model gradually removes the noise and refines its output into a trajectory.

This technique, known as Diffusion Policy, was previously introduced by researchers at MIT, Columbia University, and the Toyota Research Institute. PoCo builds off this Diffusion Policy work.

The team trains each diffusion model with a different type of dataset, such as one with human video demonstrations and another gleaned from teleoperation of a robotic arm.

Then the researchers perform a weighted combination of the individual policies learned by all the diffusion models, iteratively refining the output so the combined policy satisfies the objectives of each individual policy.

Greater than the sum of its parts

“One of the benefits of this approach is that we can combine policies to get the best of both worlds. For instance, a policy trained on real-world data might be able to achieve more dexterity, while a policy trained on simulation might be able to achieve more generalization,” Wang says.

Animation of robot arm using a spatula to lift toy pancake
With policy composition, researchers are able to combine datasets from multiple sources so they can teach a robot to effectively use a wide range of tools, like a hammer, screwdriver, or this spatula.
Image: Courtesy of the researchers

Because the policies are trained separately, one could mix and match diffusion policies to achieve better results for a certain task. A user could also add data in a new modality or domain by training an additional Diffusion Policy with that dataset, rather than starting the entire process from scratch.

Animation of robot arm using toy hammer as objects are being placed randomly next around it.
The policy composition technique the researchers developed can be used to effectively teach a robot to use tools even when objects are placed around it to try and distract it from its task, as seen here.
Image: Courtesy of the researchers

The researchers tested PoCo in simulation and on real robotic arms that performed a variety of tools tasks, such as using a hammer to pound a nail and flipping an object with a spatula. PoCo led to a 20 percent improvement in task performance compared to baseline methods.

“The striking thing was that when we finished tuning and visualized it, we can clearly see that the composed trajectory looks much better than either one of them individually,” Wang says.

In the future, the researchers want to apply this technique to long-horizon tasks where a robot would pick up one tool, use it, then switch to another tool. They also want to incorporate larger robotics datasets to improve performance.

“We will need all three kinds of data to succeed for robotics: internet data, simulation data, and real robot data. How to combine them effectively will be the million-dollar question. PoCo is a solid step on the right track,” says Jim Fan, senior research scientist at NVIDIA and leader of the AI Agents Initiative, who was not involved with this work.

This research is funded, in part, by Amazon, the Singapore Defense Science and Technology Agency, the U.S. National Science Foundation, and the Toyota Research Institute.

Read the full article

Boosting Manufacturing Profits: The Impact Of Generative AI

Boosting Manufacturing Profits: The Impact Of Generative AI

By Alice Globus

Managing cost constraints while improving output is always top of mind in the manufacturing sector. Today, corporate leadership is looking to enhance bottom-line efficiencies through the adoption of generative artificial intelligence (AI).

AI has already proven itself effective in other tech-heavy domains, such as software development, and it is one of the most viable solutions for manufacturers navigating increasingly challenging circumstances and greater pressure to achieve more with less. Today, manufacturers continue to grapple with many challenges brought about by the pandemic, along with new obstacles, such as U.S. and China relations.

For instance, while supply chain challenges have somewhat eased, the cracks exposed by the pandemic have necessitated manufacturers to enhance resilience, flexibility and reliability. Further, just as raw material prices appeared to be returning to pre-pandemic levels, new disruptions, such as trade restrictions with China, rising interest rates and inflation, have kept the cost of vital supplies elevated.

Although these disruptions may be temporary, they could be indicative of a lasting pattern. In fact, the Boston Consulting Group warns that manufacturers should brace for more “frequent and persistent” supply chain disruptions resulting from political instability, global warming and cross-sector impacts.

In addition to these supply-side challenges, recent data from The Institute for Supply Management shows that demand in the American manufacturing sector has been declining monthly since the fall of 2023. Currently, the manufacturing industry is experiencing a two-and-a-half-year low in new orders.

Despite the myriad challenges, there are also abundant opportunities thanks to significant advances in manufacturing technology, particularly AI. By harnessing a broader array of data and employing powerful machine learning and AI tools capable of identifying patterns and making predictions, manufacturers can realize various potential benefits, including:

Reducing downtime and defects through predictive maintenance.

• Addressing labor challenges through greater automation.

• Improving quality control through automated inspection.

• Enhancing supply chain optimization and flexibility.

• Improving demand forecasting.

• Enhancing long-term and real-time decision-making.

• Reducing delays and inefficiencies through better production scheduling.

• Lessening waste through green manufacturing practices.

• Improving worker safety.

By embracing AI solutions, manufacturers can identify inefficiencies, vulnerabilities, waste sources and defects more effectively. They can also forecast aspects ranging from supply chain costs to consumer demand to equipment defects more accurately.

In fact, American factories implementing sophisticated AI technologies can achieve an estimated 15% reduction in greenhouse gas emissions, along with a corresponding improvement in maintenance costs by 25%. Furthermore, McKinsey & Company estimates that predictive maintenance can reduce machine downtime by 30%-50% while extending machine life by 20%-40%.

Fully leveraging AI and enjoying its benefits, however, is not without challenges. McKinsey & Company also notes that manufacturers currently collect data from a wide range of sources, resulting in often unusable data due to quality concerns. To harness the most sophisticated AI models, organizations must address data quality issues such as broken sensors, incomplete data mappings, incompatible systems, slow access speeds and insufficient understanding of existing sources. These challenges are further compounded by the demanding environment in which cameras, sensors, machines and other data sources operate.

Improving data quality at scale, coupled with AI algorithms that can put them to use, is among the most effective solutions for enhancing the bottom line in the face of mounting challenges and constraints in the manufacturing sector. Despite the challenges of evolving from legacy systems to those that are purpose-built for the era of AI (and automation is no small feat), it is a necessary transition for improving the bottom line in the face of a mountain of challenges.

Process By Process

Small language models (SMLs) have the potential to ease some of the data challenges encountered within manufacturing by offering efficient and adaptable solutions. These models can analyze and interpret data from various sources, including sensor readings, production logs and maintenance records that provide valuable insights into operational inefficiencies, equipment failures and supply chain disruptions. Through natural language processing (NLP) capabilities, SLMs can parse unstructured data, identify patterns and extract meaningful information, which enhances data quality and usability.

Additionally, SLMs’ compact size enables the deployment of edge devices within manufacturing facilities and facilitates real-time analysis and decision-making without relying heavily on centralized computing resources. By leveraging SLMs, manufacturers can streamline data management processes, improve operational efficiency and ultimately enhance their competitive edge in an increasingly complex industrial landscape.

Read the full article

Experts See Generative AI’s Potential to Transform Public Health

Experts See Generative AI’s Potential to Transform Public Health

By Mark Hagland

A team of healthcare leaders sees major potential in leveraging generative AI for public health.

6662264c05572cc7441c8ced Dreamstime Xxl 7974598

Even as concerns continue over the sub-optimal use of artificial intelligence (AI), particularly that of generative AI, in healthcare, some healthcare leaders and researchers see tremendous opportunities for transforming the United States’ public health infrastructure using generative AI tools.

In that regard, a team of experts—Monica Bharel, M.D., M.P.H., John Auerbach, Von Guyen, and Karen B. DeSalvo—has authored a commentary article published in the June issue of Health Affairs, entitled, “Transforming Public Health Practice With Generative Artificial Intelligence,” that gets to the heart of those experts’ optimism around generative AI. The theme of the June issue of the journal is “Reimagining Public Health.”

Monica Bharel, M.D., M.P.H., is clinical lead for public sector health at Google; John Auerbach is senior vice president for public health at the ICF consulting firm; Von Guyen, M.D., is clinical lead for population health at Google; and Karen B. DeSalvo, M.D., is chief health officer at Google (and was National Coordinator for Health IT from January 2014 through August 2016).

In the commentary, the authors write that “Public health is what a nation does together as a society to create and ensure the conditions in which everyone can be healthy. Periodically, circumstances call for a dramatic shift in the ways in which optimal health is promoted. Several years ago, coinciding with a renewed focus on the social and environmental determinants of health, a shift occurred to expand traditional public health agency programs to institutionalize greater cross-sector collaboration and communitywide approaches to reach whole populations and address health inequities. Emerging around this time, the Public Health 3.0 framework was intended to expand the scope of practice for public health. One important dimension of this framework calls on the public and private sectors to work together to make real-time, geographically granular data more widely available.3 Although the field has been evolving toward a 3.0 model, using data-driven approaches to improve health outcomes, the COVID-19 pandemic revealed the need to upgrade data systems and other technological capabilities, compelling public health to evolve even further,” they write.

What’s more, they state, “More specifically, to fully realize Public Health 3.0, public health departments need to explore ways to integrate technology and new generative artificial intelligence (AI) capabilities that have garnered public attention during the past two years. We believe that generative AI will provide transformative opportunities for public health officials to approach their work. In this Commentary, we explore current uses for AI and examine the advanced capabilities of generative AI, demonstrating how they can serve as a catalyst for advancing public health. We discuss ways in which generative AI can support public health practice, including core workflows, recognizing that new technologies are most effective when they can help staff members even in the smallest local health departments. Finally, we review the challenges and risks associated with these new technologies.”

The article’s authors go through and summarize some uses of AI technology that are already very actively being pursued, including the development of large language models for predictive purposes. When it comes to generative AI, they see three primary areas of promise: “public communication,” “organizational performance,” and “novel insights.”

With regard to public communication, the article’s authors note that “Generative AI has the potential to offer a more personalized experience” to members of the public. “Some experts have argued for the importance of precision public health, and generative AI will provide tools to meet the specific information needs of individuals and the public at large. Generative AI tools can more easily create materials at multiple literacy levels and in a range of languages spoken by communities. Materials can also be personalized to suit different geographic locations and cultural factors. Content can be shared across multiple media formats, including text, audio, and images. Image generators such as Image FX and DALL-E use language prompts to generate images.25,26 These could be used by public health officials to rapidly produce visuals to aid communication efforts,” they write.

Indeed, they state, “For instance, COVID-19-era images depicting social distancing, washing hands, and wearing a face mask are ingrained in the minds of billions of people around the world, crossing all international boundaries. These images were, arguably, more universally effective at communicating important health information than text or speeches in specific languages. With text-to-image AI image generators, the barrier to creating and testing visual resources has fallen dramatically for the average health department.”

When it comes to organizational performance, the article’s authors note that “Public health uses a bureaucratic system of checks and balances and public accountability based on specific regulations and laws. This system offers transparency to the public, but it can lead to extensive clerical burden for staff at all levels. Generative AI has the potential to reduce this clerical burden by summarizing and automating administrative tasks, organizing large amounts of data, and providing analytic support to free up valuable time for staff to engage in work at the top of their professional capacity.” Generative AI tools based on large language models, they argue, “can empower the analyst by searching, extracting, and summarizing information found within large amounts of written text. Removing this clerical burden will allow the analyst to spend more time reviewing the computer-generated summary; analyzing previous health policy proposals; and designing a more thoughtful, strategic recommendation to leadership,” among other improvements.

As for the third category, “novel insights,” the authors state that “Generative AI can facilitate advanced analytics to inform new interventions to address complex health issues. For example, unstructured notes on medical examiners’ case reports, social media activity, and news reports are all potential useful text signals that are currently inaccessible at a population level. New multimodal models are emerging that will be able to integrate additional data modalities such as imaging, genomics, environmental, and geographic data into models.”

Ultimately, they conclude, “Taken together, these applications have the potential to revolutionize health by making it easier to communicate with the public, increase organizational performance, and generate novel insights into complex health problems.”

Read the full article

 

 

Generative AI provides a toolkit for decarbonization

Generative AI provides a toolkit for decarbonization

By Joseph Webster and Shaheer Hussam

Rapidly improving artificial intelligence (AI) capabilities will help accelerate the energy transition. Both established and emergent AI capabilities—such as large language models (LLMs)—can be applied to an array of strategic, technical, financial, and policy challenges posed by decarbonization. It is critical for energy transition stakeholders to monitor, understand, and carefully apply these capabilities to their unique decarbonization challenges, while also addressing the risks involved.

The most consequential new class of AI, generative AI, is able to analyze and create text, audio, code, and even molecular design—doing so faster and often with higher quality than human-created counterparts. Generative AI uses extraordinary volumes of training data and novel data-processing mechanisms which require unprecedented computational power. Data center load growth, driven by a range of factors, is forcing utilities across the United States and Europe to revisit system planning needs. Indeed, this added demand is—in some regions—delaying the retirement of coal-fired power plants. To ensure that climate targets are met, data center growth must coincide with transmission upgrades, energy efficiency improvements, and new low-carbon generation capacity. More broadly, policymakers must also consider how to harness the potential from generative AI while managing complex uncertainties, from inaccurate outputs and data leakage to AI-enabled cyberattacks on critical infrastructure. The deployment of generative AI will require rigorous human oversight, particularly in the early stages.

Given the capabilities of generative AI, integration into organizational workflows can help energy stakeholders in multiple ways—for example, lower regulatory compliance costs, consider strategic planning options, and evaluate the financial risk around their low-carbon investments, among others.

1. Strategic planning

Recent demonstrations of generative AI capabilities are impressive. Generative AI can already outline, summarize, and draft documents cheaper and faster than many humans. It can also help humans conduct strategic tasks more effectively. A study by Harvard Business School examined the effects of GPT-4—the model behind ChatGPT—on knowledge workers’ productivity, finding that GPT-4 significantly improved workers’ abilities to generate effective ideas and develop implementation plans. Another study from University College London found that a collection of LLMs could give strategic recommendations at a comparable level to human experts. As strategic planning use cases are systemic and across industries, improvements in productivity would apply across the decarbonization value chain.

2. Regulatory compliance

Some generative AI use cases will directly enhance clean energy project developers’ ability to manage cumbersome regulatory processes. As generative AI capabilities are integrated into institutional workflows, they will assist on tasks ranging from simple emails to complex, costly, and time-consuming regulatory processes. The Pacific Northwest National Laboratory, as part of its PolicyAI, initiative, recently found that LLMs could streamline the public comment-review process under the National Environmental Policy Act (NEPA), which is burdensome for many renewables firms.

Importantly, generative AI may aid regulators by accelerating reviews of a variety of environmental impact studies. For instance, after New York State attempted to ease traffic and pollution by passing traffic congestion pricing, an exhaustive environmental review took five years and more than 4,000 pages of analysis. By streamlining portions of these document-intensive regulatory tasks, generative AI can speed up environmental reviews, giving infrastructure projects a quicker go/no-go decision.

3. Decarbonization investment analytics

A range of AI tools, using both existing techniques and generative AI, are being developed to assist with financial and economic modeling, a critical but resource-intensive task for renewable energy projects. While still at the early stages, generative AI tools may be able to partially or even fully build financial models or propose complex scenario plans. In addition, AI is already being used to enhance corporate due diligence by detecting anomalies in financial statements, summarizing earnings call transcripts, or rapidly analyzing trade press. These capabilities will continue to assist both investors and corporate mergers-and-acquisitions teams in their decarbonization investments.

4. Energy asset management

Financial and economic modeling tools overlap with another essential aspect of decarbonization: advanced energy asset management. Currently, communications with energy asset field operators are typically executed via middle management and dashboards with both planned and ad hoc analytics. Generative AI may enable more simplified analytics and communication with the workers physically assessing and repairing assets. At the energy asset management level, generative AI tools could deliver improvements in compiling, summarizing, and communicating asset performance in a customized manner for financial managers.

5. Wildfire risk assessment

In parallel to generative AI, another area of quiet yet significant advancement has been machine-learning (ML) models for weather forecasting, which have produced some extraordinary results. Further advances in weather forecasting could help mitigate the climate change-driven fire season. Wildfires themselves exacerbate the climate crisis—global fires produce emissions of about 2 gigatons of carbon dioxide equivalent per year, equal to 4 percent of total global emissions. These fires can also force large populations indoors for weeks due to health risks and poor air quality. Further investment in AI/ML-based modeling could help manage these risks by predicting the probable location and magnitude of potential wildfires and improving real-time surveillance of smoke, enabling firefighters to combat the over 80,000 wildfires that occur in the United States alone every year.

Despite the current AI hype cycle and the early-stage risks around generative AI, improving the broad range of AI models will be integral to developing a low-carbon economy. The magnitude and pace will be difficult to predict, as models are integrated into institutional workflows. Human oversight, particularly around critical infrastructure, must remain comprehensive. If managed appropriately, these emergent capabilities will yield important advances in regulatory analysis, environmental management, strategic planning, and an array of challenges essential to achieving net-zero emissions.

Read the full article

Open source, open risks: The growing dangers of unregulated generative AI

Open source, open risks: The growing dangers of unregulated generative AI

By Charles Owen-Jackson

While mainstream generative AI models have built-in safety barriers, open-source alternatives have no such restrictions. Here’s what that means for cyber crime.

There’s little doubt that open-source is the future of software. According to the 2024 State of Open Source Report, over two-thirds of businesses increased their use of open-source software in the last year.

Generative AI is no exception. The number of developers contributing to open-source projects on GitHub and other platforms is soaring. Organizations are investing billions in generative AI across a vast range of use cases, from customer service chatbots to code generation. Many of them are either building proprietary AI models from the ground up or on the back of open-source projects.

But legitimate businesses aren’t the only ones investing in generative AI. It’s also a veritable goldmine for malicious actors, from rogue states bent on proliferating misinformation among their rivals to cyber criminals developing malicious code or targeted phishing scams.

Tearing down the guard rails

For now, one of the few things holding malicious actors back is the guardrails developers put in place to protect their AI models against misuse. ChatGPT won’t knowingly generate a phishing email, and Midjourney won’t create abusive images. However, these models belong to entirely closed-source ecosystems, where the developers behind them have the power to dictate what they can and cannot be used for.

It took just two months from its public release for ChatGPT to reach 100 million users. Since then, countless thousands of users have tried to break through its guardrails and ‘jailbreak’ it to do whatever they want — with varying degrees of success.

The unstoppable rise of open-source models will render these guardrails obsolete anyway. While performance has typically lagged behind that of closed-source models, there’s no doubt open-source models will improve. The reason is simple — developers can use whichever data they like to train them. On the positive side, this can promote transparency and competition while supporting the democratization of AI — instead of leaving it solely in the hands of big corporations and regulators.

However, without safeguards, generative AI is the next frontier in cyber crime. Rogue AIs like FraudGPT and WormGPT are widely available on dark web markets. Both are based on the open-source large language model (LLM) GPT-J developed by EleutherAI in 2021.

Malicious actors are also using open-source image synthesizers like Stable Diffusion to build specialized models capable of generating abusive content. AI-generated video content is just around the corner. Its capabilities are currently limited only by the availability of high-performance open-source models and the considerable computing power required to run them.

What does this mean for businesses?

It might be tempting to dismiss these issues as external threats that any sufficiently trained team should be adequately equipped to handle. But as more organizations invest in building proprietary generative AI models, they also risk expanding their internal attack surfaces.

One of the biggest sources of threat in model development is the training process itself. For example, if there’s any confidential, copyrighted or incorrect data in the training data set, it might resurface later on in response to a prompt. This could be due to an oversight on the part of the development team or due to a deliberate data poisoning attack by a malicious actor.

Prompt injection attacks are another source of risk, which involves tricking or jailbreaking a model into generating content that goes against the vendor’s terms of use. That’s a risk facing every generative AI model, but the risks are arguably greater in open-source environments lacking sufficient oversight. Once AI tools are open-sourced, the organizations they originate from lose control over the development and use of the technology.

The easiest way to understand the threats posed by unregulated AI is to ask the closed-source ones to misbehave. Under most circumstances, they’ll refuse to cooperate, but as numerous cases have demonstrated, all it typically takes is some creative prompting and trial and error. However, you won’t run into any such restrictions with open-source AI systems developed by organizations like Stability AI, EleutherAI or Hugging Face — or, for that matter, a proprietary system you’re building in-house.

A threat and a vital tool

Ultimately, the threat of open-source AI models lies in just how open they are to misuse. While advancing democratization in model development is itself a noble goal, the threat is only going to evolve and grow and businesses can’t expect to count on regulators to keep up. That’s why AI itself has also become a vital tool in the cybersecurity professional’s arsenal. To understand why, read our guide on AI cybersecurity.

Read the full article

 

Apple Intelligence: The AI features coming to your iPhone

Apple Intelligence: The AI features coming to your iPhone

By Tim Biggs

Apple is adding generative AI tools to its phones and computers that can rewrite and summarise text, create “cartoony” images, edit photos and more, as well as an optional integration with OpenAI’s ChatGPT.

The collection of tools, which the company calls Apple Intelligence, will arrive later this year for iOS 18 and MacOS Sequoia, which were unveiled at the annual Worldwide Developer Conference on Tuesday morning (AEST). They also include an upgraded and more natural Siri assistant, which now understands text prompts and can respond with step-by-step instructions.

Many of the iOS tools, which will work on iPhone 15 Pro or later, are similar to those already available on Google and Samsung phones or through Microsoft apps such as Office. But Apple chief executive Tim Cook said the company was doing things differently.

“Apple Intelligence will transform what users can do with our products, and what our products can do for our users,” he said, in announcing the tools.

“Our unique approach combines generative AI with a user’s personal context to deliver truly helpful intelligence. And it can access that information in a completely private and secure way to help users do the things that matter most to them.”

Apple AI features will be available across Mac, iPad and iPhone.
Apple AI features will be available across Mac, iPad and iPhone.

The new tools include the ability to highlight any text and have it be rewritten in a different style, or summarise text from web pages or any app. Users will also be able to record audio and have it transcribed and summarised automatically, from either the Notes app or while on a phone call. For the latter, the other participants in the call will be alerted to the fact they’re being recorded.

Image generation will work in several applications but will also have its own dedicated app called Image Playground. Users can add prompts and descriptions to create images instantly, and can even choose people to include from their Photos library, but the output is always in a “cartoony” style rather than a realistic one.

In Messages the system will suggest custom stickers based on conversations, while in Notes a scribbled picture can be transformed into a more defined image. Users can also create custom emojis using text prompts.

In Photos, a new clean-up tool can remove distracting background elements with a tap, while users can also surface specific images by describing them.

Bringing all of the Apple Intelligence tools together is a redesigned Siri, which can be interacted with using voice or text. Apple said Siri would be able to look at a user’s screen for context or dig through their data, to accomplish complex tasks such as “Play that podcast that Jamie recommended” or “Send the photos from the barbecue on Saturday to Malia.”

Siri and Apple Intelligence can use Chat-GPT for some requests, such as generating or refining text.
Siri and Apple Intelligence can use Chat-GPT for some requests, such as generating or refining text.

Apple stressed that it was taking a privacy-first approach to AI, with as much data processed on-device as possible. Any cloud processing is done in Apple’s custom secure servers, which limit personal data collected. However, the system can also use external AI services, the first of which is OpenAI’s ChatGPT.

When a user asks for something that’s best handled by ChatGPT, the device will ask permission to share the request, and OpenAI will not log the user’s IP address or store their data unless they’re a paying OpenAI customer who has linked their account. The ChatGPT integration is free to use on Apple devices.

Vision Pro and other updates

Also at the conference, Apple announced that its Vision Pro headset would arrive in Australia on July 12, and would start at $6000. The so-called spatial computer is also getting its first annual software update with VisionOS 2, including the ability to turn regular photos into 3D spatial photos, and a new extremely wide-screen virtual display for use with Mac computers.

Apple’s Vision Pro headset is coming to Australia in July.
Apple’s Vision Pro headset is coming to Australia in July.

Aside from Apple Intelligence, iOS 18 was shown with a redesigned Home Screen that allows more customisable app layouts and the ability to tint app icons with any colour. An AI system will now also identify notifications, messages and emails it thinks are high priority to emphasise them above others.

The iPhone will also get the much-requested ability to send and receive RCS messages, which should mean better multimedia support between Apple and Android devices. A new Passwords app offers an Apple-focused option for a password manager, while users of recent AirPods models will be able to shake or nod their head to interact with Siri.

iPadOS is getting many of the same features, but a special mention was given to a new calculator app. Not only can it function like a regular calculator, but it will also work out and respond to sums you write by hand, any time you create an equals sign.

Apple TV is taking a cue from Amazon with a new feature called InSight, which shows you which actors are in a scene when you pause a show or movie. It will also display what song is playing, and let you add it to your Apple Music library.

Finally, a new continuity feature in MacOS lets you control your iPhone from your computer. While the phone stays locked, users are able to navigate the device and even use apps by interacting with a virtual version of their phone on their computer screen.

Read the full article