In its 13th year, latest report also notes proliferation of ideas and tools to improve accessibility and avoid ballooning cloud complexity
Thoughtworks (NASDAQ: TWKS), a global technology consultancy that integrates strategy, design and engineering to drive digital innovation, today released Volume 28 of the Technology Radar, a biannual report informed by Thoughtworks’ observations, conversations and frontline experiences solving its global clients’ toughest business challenges. With artificial intelligence (AI) becoming even more accessible and embedded in the business, the report states that implementing AI with robust engineering practice leads to the most effective delivery of value to customers.
With analytics and AI, having enough data and code is no longer the problem. Instead, the focus should be on having high-quality code and data to create models that properly respond to changes in environmental factors and do not drift over time. Fortunately, tooling in this area is expanding with test-driven transformations, data sanity tests and data model testing that strengthen the data pipelines that power analytical systems. Incorporating best practices in model validation and quality assurance are also crucial in tackling biases and ensuring ethical ML systems with equitable outcomes.
“Most businesses are already experimenting with analytics and AI. We see the maturing of tools as further evidence of the mainstreaming of these technologies. Integration of AI with good engineering practices leads to more responsible, data-driven solutions that cater to a diverse user base,” said Dr. Rebecca Parsons, chief technology officer at Thoughtworks. “This is in contrast to generative AI, where we’re advising businesses to proceed cautiously and take care to avoid inappropriate uses that can lead to reputational and security risks.”
Highlighted themes in Technology Radar Vol. 28 include:
- The meteoric rise of practical AI: In recent months, tools like ChatGPT have completely reoriented both the technology industry’s and the wider world’s understanding of what’s practically possible with AI. We urge users to explore these opportunities while also remaining sensitive to their limitations and risks.
- Accessible accessibility: Accessibility is not a novel concern, but the proliferation of ideas and tools in this space means product and engineering teams can address it fairly easily. There are now few excuses for failing to take accessibility seriously and embedding it into the things you build.
- Lambda quicksand: Serverless functions can be extremely useful, but they can also encourage bad habits and lead to poor architectural decisions which increase complexity. To get the most from serverless, be clear about where you use it and be aware of the potential consequences.
- Engineering rigor meets analytics and AI: Without high-quality data, teams implementing AI risk compromising accuracy and, by extension, user trust. This can only be done effectively through robust engineering practices such as model validation, continuous integration and monitoring. Fortunately, the tooling here is expanding — we encourage teams to leverage the growing ecosystem.
- To declare or program?: Choosing between declarative specifications or general-purpose programming languages for certain tasks is an important ongoing consideration for software developers. There’s rarely an obvious answer, which means it’s essential to always reflect on the benefits and risks when faced with a new context.
Visit AITechPark for cutting-edge Tech Trends around AI, ML, Cybersecurity, along with AITech News, and timely updates from industry professionals!
