Archive 25.01.2025

This month, Microsoft refined its Copilot pricing for Office 365, and you get a lot for your extra $30 a month. But this also reminds me of a product called Crosstalk that I followed back in the early 90s, right […]

The post Why I Think Copilot Means the End of Office as We Know It appeared first on TechSpective.

When designing a robot, such as Boston Dynamics' anthropomorphic robot Atlas, which appears exercising and sorting boxes, fiducial markers are the guides that help them move, detect objects and determine their exact location. It is a machine vision tool that is used to estimate objects' positions. At first glance they are flat, high-contrast black and white square codes, roughly resembling the QR marking system, but with an advantage: they can be detected at much greater distances.

Researchers at the Technical University of Darmstadt and the Helmholtz Center Dresden-Rossendorf have developed flexible robot wings that are moved by magnetic fields. Inspired by the efficiency and adaptability of the wings of the monarch butterfly, they enable precise movements without electronics or batteries.

This year, we dive deeper into the trends driving AIoT, showing how they are reshaping industries and fostering a more efficient, secure, and sustainable future.

Claire chatted to Didem Gurdur Broo from Uppsala University about how to shape the future of robotics, autonomous vehicles, and industrial automation.

Didem Gurdur Broo is an Assistant Professor and Associate Senior Lecturer at the Department of Information Technology at Uppsala University. She leads the Cyber-physical Systems Lab, directing research on intelligent systems like collaborative robots, autonomous vehicles, and smart cities. Didem is a computer scientist with a PhD in mechatronics, which can give you an idea about how much she loves to talk about the future and emerging technologies. She dreams a better world and actively works on improving inequalities regardless of their nature.

We humans excel at generalization. If you taught a toddler to identify the color red by showing her a red ball, a red truck and a red rose, she will most likely correctly identify the color of a tomato, even if it is the first time she sees one.

While we move through 2025, pen and paper remain as familiar on construction sites as ever, and many construction managers still rely on intuition and experience over the promises of technology or robotic innovation.

Recent advances in robotics and artificial intelligence (AI) are enabling the development of a wide range of systems with unique characteristics designed for varying real-world applications. These include robots that can engage in activities traditionally only completed by humans, such as sketching, painting and even hand-writing documents.

As part of the ongoing war in Ukraine, one night in late November, Russia sent a swarm of 188 drones to attack Ukrainian infrastructure like electrical utilities, as well as residential areas, according to news reports. Ukrainian forces said they shot down 76 drones, but the damage was still extensive. Those kinds of attacks are continuing almost daily now.

Generative AI holds incredible promise, but its potential is often blocked by poor app experiences. 

AI leaders aren’t just grappling with model performance — they’re contending with the practical realities of turning generative AI into user-friendly applications that deliver measurable enterprise value.

Infrastructure demands, unclear output expectations, and complex prototyping processes stall progress and frustrate teams.

The rapid pace of AI innovation has also introduced a growing patchwork of tools and processes, forcing teams to spend time on integration and basic functionality instead of delivering meaningful business solutions.

This blog explores why AI teams encounter these hurdles and offers actionable solutions to overcome them.

What stands in the way of effective generative AI apps?

While teams move quickly on technical advancements, they often face significant barriers to delivering usable, effective business applications: 

• Technology complexity: Building the infrastructure to support generative AI apps — from vector databases to Large Language Model (LLM) orchestration — requires deep technical expertise that most organizations lack. Choosing the right LLM for specific business needs adds another layer of complexity.

• Unclear objectives: Generative AI’s unpredictability makes it hard to define clear, business-aligned objectives. Teams often struggle to connect AI capabilities into solutions that meet real-world needs and expectations.

• Talent and expertise: Generative AI moves fast, but skilled talent to develop, manage, and govern these applications is in short supply. Many organizations rely on a patchwork of roles to fill gaps, increasing risk and slowing progress.

• Collaboration gaps: Misalignment between technical teams and business stakeholders often results in generative AI apps that miss expectations — both in what they deliver and how users consume them.

• Prototyping barriers: Prototyping generative AI apps is slow and resource-intensive. Teams struggle to test user interactions, refine interfaces, and validate outputs efficiently, delaying progress and limiting innovation.

• Hosting difficulties: High computational demands, integration complexities, and unpredictable outcomes often make deployment challenging. Success requires not only cross-functional collaboration but also robust orchestration and tools that can adapt to evolving needs. Without workflows that unite processes, teams are left managing disconnected systems, further delaying innovation.

The result? A fractured, inefficient development process that undermines generative AI’s transformative potential.

Despite these app experience hurdles, some organizations have navigated this landscape successfully. 

For example, after carefully evaluating its needs and capabilities, The New Zealand Post — a 180-year-old institution — integrated generative AI into its operations, reducing customer calls by 33%.

Their success highlights the importance of aligning generative AI initiatives with business goals and equipping teams with flexible tools to adapt quickly.

Turn generative AI challenges into opportunities

Generative AI success depends on more than just technology — it requires strategic alignment and robust execution. Even with the best intentions, organizations can easily misstep.

Overlook ethical considerations, mismanage model outputs, or rely on flawed data, and small mistakes quickly snowball into costly setbacks.

AI leaders must also contend with rapidly evolving technologies, skill gaps, and mounting demands from stakeholders, all while ensuring their models are secure, compliant, and reliably perform in real-world scenarios.

Here are six strategies to keep your initiatives on track:

1. Business alignment and needs assessment: Anchor your AI initiatives to your organization’s mission, vision, and strategic objectives to ensure meaningful impact.
   
   
2. AI technology readiness: Assess your infrastructure and tools. Does your organization have the tech, hardware, networking, and storage to support generative AI implementation? Do you have tools that enable seamless orchestration and collaboration, allowing teams to deploy and refine models quickly?
   
   
3. AI security and governance: Embed ethics, security, and compliance into your AI initiatives. Establish processes for ongoing monitoring, maintenance, and optimization to mitigate risks and ensure accountability.
   
   
4. Change management and training: Foster a culture of innovation by building skills, delivering targeted training, and assessing readiness across your organization.
   
   
5. Scaling and continuous improvement: Identify new use cases, measure and communicate AI impact, and continually refine your AI strategy to maximize ROI. Focus on reducing time-to-value by adopting workflows that are adaptable to your specific business needs, ensuring that AI delivers real, measurable outcomes.
   

Generative AI isn’t an industry secret — it’s transforming businesses across sectors, driving innovation, efficiency, and creativity.

Yet, according to our Unmet AI Needs survey, 66% of respondents cited difficulties in implementing and hosting generative AI applications. But with the right strategy, businesses in virtually every industry can gain a competitive edge and tap into AI’s full potential. 

Lead the way to generative AI success

AI leaders hold the key to overcoming the challenges of implementing and hosting generative AI applications. By setting clear goals, streamlining workflows, fostering collaboration, and investing in scalable solutions, they can pave the way for success.

To achieve this, it’s critical to move beyond the chaos of disconnected tools and processes. AI leaders who unify their models, teams, and workflows gain a strategic advantage, enabling them to adapt quickly to changing demands while ensuring security and compliance.

Equipping teams with the right tools, targeted training, and a culture of experimentation transforms generative AI from a daunting initiative into a powerful competitive advantage.

Want to dive deeper into the gaps teams face with developing, delivering, and governing AI? Explore  our Unmet AI Needs report for actionable insights and strategies.

The post Solving the generative AI app experience challenge appeared first on DataRobot.

The culmination of the challenge will be in-person presentations to industry, investors, and community members at the Robotics Summit & Expo in Boston from April 30 to May 1st.

Experts highlight the need for a clear framework when it comes to AI research, given the rapid adoption of artificial intelligence by children and adolescents using digital devices to access the internet and social media.

Research in the International Journal of Computational Vision and Robotics could lead to faster and more accurate robots for high-precision tasks in factories.

DEEP Robotics develops a range of advanced robotic platforms, each designed for specific real-world applications across industrial, public service, and research sectors.

Lite3 is an agile quadruped platform ideal for education, research, and light industrial tasks, offering versatility for developing AI and robotics solutions.
Lite3 supports open source, users can develop advanced perception capabilities such as autonomous navigation, automatic obstacle avoidance, visual localization, environment reconstruction, customizable API for robotics development and AI training

Image credit: Deep Robotics – deeprobotics.cn

X30, a robust quadruped, is tailored for inspection, security, and autonomous navigation in hazardous environments such as industrial sites and power plants.
X30 robot dog conducts autonomous inspection day and night in any weather stably, the operating temperature range of X30 has been extended to between minus 20°C and plus 55°C, the load capacity can be up to 85KG

Image credit: Deep Robotics – deeprobotics.cn

DEEP Robotics Lynx, an off-road wheeled quadruped, is built for outdoor exploration, search-and-rescue missions, providing unmatched mobility on uneven terrain.
DEEP Robotics Lynx all-terrain robot features an agile design and powerful multi-terrain adaptability, combined with a distinctive wheel-legged movement system and AI driven, striking an ideal balance between speed and agility

Image credit: Deep Robotics – deeprobotics.cn

DR01 humanoid robot showcases advanced locomotion for dynamic human-like motion, contributing to research in human-robot interaction and service robotics for task automation.
DR01 humanoid robot boasts highly flexible movement capabilities, adapts to complex environments, and integrates sensing/perception abilities and powerful autonomous learning capabilities

Image credit: Deep Robotics – deeprobotics.cn

Various Challenges for the systems above: Enhancing autonomy, improving power efficiency, and refining adaptability to diverse environments.

Future Goals for the systems:
Integrating AI-driven perception systems for smarter navigation.
Expanding industrial, public safety, and service applications.
Developing robots for seamless, collaborative human-robot interactions with superior performance in real-world scenarios.

Videos:
https://www.youtube.com/watch?v=zxGwOEYYFVo
https://www.youtube.com/watch?v=NNkxlKLMoMM&t=19s
https://www.youtube.com/watch?v=iL833P0Vino
https://www.youtube.com/watch?v=lCFyfh3mLOQ

The content and media above is provided to us by Deep Robotics. www.deeprobotics.cn

Scientists have developed a computing chip that can learn, correct errors, and process AI tasks.