Risk is all about context. Actually, one in every of the largest risks is failing to acknowledge or understand context: That’s why it’s essential begin there when evaluating risk.

This is especially vital when it comes to status. Think, as an example, about your customers and their expectations. How might they feel about interacting with an AI chatbot? How damaging might it’s to offer them with false or misleading information? Perhaps minor customer inconvenience is something you possibly can handle, but what if it has a major health or financial impact?

Even when implementing AI seems to make sense, there are clearly some downstream status risks that must be considered. We’ve spent years talking in regards to the importance of user experience and being customer-focused: While AI might help us here, it could also undermine those things as well.

There’s the same query to be asked about your teams. AI can have the capability to drive efficiency and make people’s work easier, but utilized in the mistaken way it could seriously disrupt existing ways of working. The industry is talking quite a bit about developer experience recently—it’s something I wrote about for this publication—and the choices organizations make about AI need to enhance the experiences of teams, not undermine them.

In the most recent edition of the Thoughtworks Technology Radar—a biannual snapshot of the software industry based on our experiences working with clients around the globe—we discuss precisely this point. We call out AI team assistants as probably the most exciting emerging areas in software engineering, but we also note that the main target must be on enabling teams, not individuals. “You need to be on the lookout for ways to create AI team assistants to assist create the ‘10x team,’ versus a bunch of siloed AI-assisted 10x engineers,” we are saying in the most recent report.

Failing to heed the working context of your teams could cause significant reputational damage. Some bullish organizations might see this as part and parcel of innovation—it’s not. It’s showing potential employees—particularly highly technical ones—that you just don’t really understand or care in regards to the work they do.

There are a number of tools that will be used to assist manage risk. Thoughtworks helped put together the Responsible Technology Playbook, a set of tools and techniques that organizations can use to make more responsible decisions about technology (not only AI).

Nonetheless, it’s vital to notice that managing risks—particularly those around status—requires real attention to the specifics of technology implementation. This was particularly clear in work we did with an assortment of Indian civil society organizations, developing a social welfare chatbot that residents can interact with of their native languages. The risks here weren’t unlike those discussed earlier: The context through which the chatbot was getting used (as support for accessing vital services) meant that incorrect or “hallucinated” information could stop people from getting the resources they depend upon.

This contextual awareness informed technology decisions. We implemented a version of something called retrieval-augmented generation to scale back the danger of hallucinations and improve the accuracy of the model the chatbot was running on.

Retrieval-augmented generation features on the most recent edition of the Technology Radar. It is perhaps viewed as a part of a wave of emerging techniques and tools on this space which might be helping developers tackle among the risks of AI. These range from NeMo Guardrails—an open-source tool that puts limits on chatbots to extend accuracy—to the strategy of running large language models (LLMs) locally with tools like Ollama, to make sure privacy and avoid sharing data with third parties. This wave also includes tools that aim to enhance transparency in LLMs (that are notoriously opaque), comparable to Langfuse.

It’s price stating, nevertheless, that it’s not only a matter of what you implement, but in addition what you avoid doing. That’s why, on this Radar, we caution readers in regards to the dangers of overenthusiastic LLM use and rushing to fine-tune LLMs.

A brand new wave of AI risk assessment frameworks aim to assist organizations consider risk. There may be also laws (including the AI Act in Europe) that organizations must listen to. But addressing AI risk isn’t just a matter of applying a framework and even following a static set of excellent practices. In a dynamic and changing environment, it’s about being open-minded and adaptive, paying close attention to the ways in which technology selections shape human actions and social outcomes on each a micro and macro scale.

One useful framework is Dominique Shelton Leipzig’s traffic light framework. A red light signals something prohibited—comparable to discriminatory surveillance—while a green light signals low risk and a yellow light signals caution. I like the actual fact it’s so lightweight: For practitioners, an excessive amount of legalese or documentation could make it hard to translate risk to motion.

Nonetheless, I also think it’s price flipping the framework, to see risks as embedded in contexts, not within the technologies themselves. That way, you’re not attempting to make an answer adapt to a given situation, you’re responding to a situation and addressing it because it actually exists. If organizations take that approach to AI—and, indeed, to technology typically—that can ensure they’re meeting the needs of stakeholders and keep their reputations secure.