Technology Tales

Fathom is a business management solution that integrates reporting, analysis and forecasting tools to provide clear insights into financial performance. It offers AI-generated commentary tailored to specific business contexts, enabling users to create custom management reports quickly and share results effectively.

Features include cash flow forecasting, scenario evaluation and consolidated financial reporting for groups, supported by integration with major accounting platforms. The platform is designed for businesses of all sizes, offering tools to measure key performance indicators, benchmark company performance and streamline financial planning. User testimonials highlight its value in enhancing clarity and confidence in financial decision-making, with a focus on transparency, automation and ease of use.

Learning Machines is a data science blog covering a range of topics including machine learning, statistical computing, quantitative finance and the R programming language. Recent posts explore building a transformer-based language model in R, the relationship between income and happiness, the role of artificial intelligence in academic work, regression to the mean in business contexts, Youth Bulge Theory as a lens for understanding Middle East conflict, the reliability of election polling, the distinction between weather and climate, stock market simulation using multi-agent models, trading strategy analysis and an introductory guide to R for newcomers to statistical programming.

Apache Spark is a versatile engine designed for large-scale data analytics, supporting data engineering, science and machine learning tasks across single-node systems or clusters. It processes batch and real-time streaming data using multiple programming languages, executes distributed SQL queries efficiently and enables scalable data analysis without downsampling. The platform integrates with widely used frameworks for data science, machine learning and business intelligence, while its SQL engine optimises query execution through adaptive techniques and supports both structured and unstructured data formats. Widely adopted by numerous organisations, it benefits from a large open-source community and extensive ecosystem, facilitating deployment across diverse infrastructure and storage solutions.

A comprehensive survey of over 1,500 IT executives by Cloudera reveals that whilst enterprises remain bullish about artificial intelligence investments, only 21% have achieved full AI integration into their core business processes. The primary barriers include rising costs for the compute capacity needed for model training, which jumped from 8% to 42% year-on-year, alongside challenges in accessing comprehensive organisational data across different environments.

Successful AI implementation requires a structured approach beginning with clear business objectives, followed by data unification and infrastructure development that prioritises security and governance from the outset. Early wins typically emerge from focused, ROI-driven domains such as IT helpdesk automation and DevOps assistance, where measurable improvements in operational efficiency, customer experience and productivity can be demonstrated.

Security remains paramount, with half of respondents concerned about training data leaks and unauthorised access. This necessitates governance frameworks that bring AI to data rather than moving data to AI systems. Compliance must be embedded by design rather than retrofitted, with policies applied universally across cloud and on-premises environments.

Achieving ubiquitous AI deployment depends not merely on solving technical challenges around data silos and infrastructure costs. It depends fundamentally on building trust through explainable decisions grounded in reliable, well-governed data that provides visibility into AI decision-making processes.

OpenText provides secure information management solutions designed to help organisations harness data effectively for AI-driven outcomes. By connecting data across enterprise systems, including CRM, ERP and cloud platforms, the company enables organisations to prepare data for AI, ensure compliance and automate workflows through customised tools.

With a global presence spanning over 180 countries and trusted by numerous major corporations, OpenText focuses on securing data through encryption and governance frameworks while offering a range of products and services that support analytics, cybersecurity, content management and DevOps. The company's approach integrates AI into business processes, aiming to reduce errors, enhance efficiency and deliver scalable results across industries.

Recent market swings following a warning from OpenAI's chief executive about investor overexcitement have highlighted a pattern that has repeated across centuries. Whenever a genuinely transformative technology is followed by a surge of speculative capital, the same sequence tends to unfold. It can be seen in the railway mania of the 1840s, the dot-com frenzy of the 1990s, Japan's asset bubble in the 1980s and China's managed stock market peaks. Each episode ended with a dramatic correction but left behind physical or digital infrastructure that later underpinned industrial growth.

The current wave, driven by generative AI, differs in one significant respect: information about failure rates and valuation excesses is instantly available to a broad audience. Yet the technology's rapid deployment, potential for self-improvement and network-enhanced learning could change the tempo of progress, possibly avoiding a classic boom-and-bust or alternatively creating a bubble that delivers more disruption while still rising.

Analysts note that the typical cycle for such bubbles lasts four to six years. The mid-phase of the AI surge, characterised by heavy capital outlays for data centres and silicon, is already underway. This suggests that a realignment may be imminent if monetary policy tightens or if the promised commercial returns lag behind the hype.

Many investors are tempted to chase the next headline. The experience of past bubbles, however, points to the importance of concentrating on durable infrastructure and companies that solve real needs, combined with the patience to wait for the market to correct itself. These are lessons that are easier to understand now than they were when railways or early internet stocks were first introduced.

Trustworthy and Responsible AI (SAS Institute)

This course from SAS Institute provides foundational knowledge about trust and responsibility in artificial intelligence and machine learning systems. It targets anyone involved in making business decisions based on AI or designing AI systems, regardless of their role, and requires no formal prerequisites beyond basic data literacy.

The programme covers how trustworthy AI integrates with analytics life cycles and data supply chains, with a focus on identifying and addressing unwanted biases throughout these processes. Participants learn six core principles of responsible innovation, including human-centricity, inclusivity, accountability, privacy and security, robustness and transparency, explored through practical scenarios ranging from healthcare risk models to speech recognition systems.

The curriculum examines real-world examples such as racial bias in research, mobile device encryption, cryptocurrency exchange failures and credit rating agency practices to illustrate these principles in action. Each module is designed to take under an hour and can be completed at one's own pace, making the course accessible to data consumers, IT professionals, managers, analysts, data scientists and decision-makers across a wide range of industries.

A comprehensive course explores generative artificial intelligence and its practical applications through SAS tools, covering approximately four hours of content with hands-on practice components. The programme examines various types of generative AI systems within the broader AI landscape, addressing key challenges and opportunities in developing trustworthy AI solutions.

Students learn to generate synthetic data using techniques such as Synthetic Minority Oversampling Technique and Generative Adversarial Networks, whilst exploring how large language models produce meaningful content through transformer architecture and attention mechanisms. The curriculum also includes practical instruction on using Bidirectional Encoder Representations from Transformers for content classification and implementing Retrieval Augmented Generation to enhance large language model output accuracy and relevance.

Designed for learners with an existing background in statistics and machine learning using SAS, the course takes a phased release approach with new lessons added periodically to reflect the rapidly evolving field. It covers everything from fundamental generative AI concepts to advanced implementation techniques within SAS Viya and SAS Machine Learning environments.

OpenAI has released GPT-5, their most advanced model for coding and agentic tasks, now available through their API platform in three sizes: gpt-5, gpt-5-mini and gpt-5-nano. The model achieves state-of-the-art performance across key coding benchmarks, scoring 74.9% on SWE-bench Verified and 88% on Aider polyglot. It demonstrates particular excellence in frontend development, where it outperformed OpenAI o3 in 70% of internal tests.

GPT-5 excels at collaborative coding tasks, bug fixing and handling complex codebases, with enhanced capabilities for chaining together multiple tool calls in sequence or in parallel without losing context. The model introduces new API features including adjustable verbosity levels (low, medium and high), a minimal reasoning effort option for faster responses, and custom tools that allow plaintext input instead of JSON formatting.

Beyond coding, GPT-5 shows significant improvements in instruction following, achieving 69.6% on Scale MultiChallenge, and demonstrates superior performance in long-context tasks with support for up to 400,000 total tokens. The model also exhibits substantially improved factual accuracy, making approximately 80% fewer factual errors than previous models on Long Fact and FactScore benchmarks. This makes it more suitable for high-stakes applications where correctness is essential.

Early testing partners including Cursor, Windsurf and Vercel have provided positive feedback regarding the model's intelligence, steerability and reduced error rates compared to other frontier models. These early assessments suggest the release marks a meaningful step forward for developer-focused AI tooling.

The development of Good Machine Learning Practice (GMLP) for medical device innovation is at the forefront of regulatory initiatives led by the U.S. FDA, Health Canada and the UK's Medicines and Healthcare Products Regulatory Agency. These organisations have outlined ten guiding principles aimed at promoting the safe and effective use of AI and machine learning technologies in healthcare.

Multidisciplinary expertise throughout the product lifecycle is considered essential for integrating machine learning models into clinical workflows safely and effectively whilst addressing patient needs. Equally important are representative data sets in clinical studies, independence between training and test data sets, and the selection of reference data based on the best available methods, all of which are necessary for generalising results across intended patient populations.

Appropriately tailored model design can mitigate risks such as overfitting and security vulnerabilities, with the focus extending beyond the models themselves to the performance of the human-AI team as a whole. Monitoring real-world use whilst managing retraining risks, providing users with clear and contextually relevant information, and maintaining robust software engineering and security practices are all considered imperative.

This collaborative framework aims to advance GMLP standards and regulatory guidelines by encouraging international cooperation, harmonisation and innovation in AI-powered medical technologies. Users are encouraged to engage with these developments and provide feedback through dedicated platforms.