This piece is a little different from our usual ones about potential technological applications that can benefit businesses. Instead, we are focusing on how to successfully realise said benefits by avoiding certain pitfalls, specifically concerning AI projects. Surprisingly, not all AI pitfalls are technical and many of them directly relate to business aspects of AI work. Therefore, business stakeholders can make all the difference with vital knowledge about how to avoid AI business failure.
Handling the five main business-oriented challenges for AI work:
1. Unquantified or unstated business goals
Anecdotal evidence suggests it’s increasingly common for business leaders to rush into implementing AI projects because it seems like everyone else is talking about and using AI. However, when starting AI work, businesses need to prioritise clarity on what it’s expected to achieve, expressed in measurable terms. Business cases can be risky without benchmarks and published data to predict returns on AI investments. Group AI spend with existing investments such as R&D and set “soft” business goals that you would for other projects under that category, allowing leeway for occasional failures. Alternatively, if you already have early projects in motion, use these to create your baseline of metrics for future AI business use cases.
2. Insufficient focus on business impact
The best IT projects balance business and technology activities, but this is yet to become the norm for AI business projects. This may be because AI is usually seen as a specialist technical skill that does not require involvement from business people, who are not considered to be qualified in the field. At the early stages of AI business work, it’s important to make sure business implementation is a standard part of any status reporting, particularly for senior management. Business leaders should be included in core AI teams, instead of regarded as consultants for occasional “user” input.
3. Hiring the wrong skills
A common AI pitfall is hiring skills without clarity around what those should be and why. The requirements are starkly different if you’re staffing AI projects in-house or using third parties for AI work. You’d need to hire staff with data science skills for internal project development, which is an entirely different profile to someone needed to procure and manage third party data scientists / AI consultancies. Issues intensify when there’s a lack of consensus around how to structure AI teams in the organisation. HR and procurement personnel are rarely well versed when it comes to evaluating AI candidates or suppliers. Awareness of this pitfall should make it easier to avoid costly mistakes around permanent hires by prioritising flexibility until you have a clear AI sourcing strategy in place.
4. Poorly conceptualised business questions
One of the major AI traps is expecting the technology to answer everything, but this can lead to ambiguous outcomes when businesses ask unreasonable or unrealistic questions. For example, in a scenario where customer retention is a core business issue, AI may be leveraged to accurately spot customers when they look likely to leave the store. This would correspond with the moment in time where a customer is expressing their greatest level of dissatisfaction. At this stage, attempts at intervention from engaging members of staff are doomed to fail as options to change the customer’s mind are harder and fewer than they would’ve been earlier on in the relationship. Business stakeholders are in a good position to consider how AI results will be used in business terms and can therefore help to develop better thought-out business questions, such as asking AI to uncover early signs of dissatisfaction and causation insights so that interventions have a higher probability of being successful.
5. AI for the sake of AI
This final pitfall is the easiest to spot and avoid – doing AI work primarily because of a desire to use the technology, perhaps because everyone else seems to be. It goes together with insufficient thought on the kind of problem to solve with AI and coupled together it leads to projects that are trying to solve unrealistic questions and ultimately lack business focus. To avoid this AI pitfall, reflect on whether a project is being driven by a genuine business problem or is a symptom of the desire to use the technology.
Regardless of scope, all AI projects ultimately seek success, and it is avoiding these five failures that will put you ahead of many others who are exploring AI in business.
Blockchain and cannabis are two highly disruptive trends that are fundamentally shifting how business is conducted across the…