26 . Juli 2017
Posts nach Autoren: Dr Simon Thompson, Head of Big Data and Customer Experience Practice, BT
Creating value in artificial intelligence (AI) is a challenge faced by many businesses. From proving the investment to system end of life, it’s something that’s hindering adoption amongst more traditional companies.
Something needs to change, because AI is not going away. Gartner predicts that by 2020, AI will be a top five investment priority for more than 30 per cent of CIOs, and it will be in almost every new software product by 2020 . The good news is that there have been some advances which are making it easier to implement and adopt AI in the workplace.
There have been some major changes in the data space over the last decade. In addition to the reams of new data sources within today’s digital enterprises, we now also have much more capability to store and process ‘big’ data.
Then we have the next generation of networks which are making it faster to move data around. This means that we can instrument every aspect of business operations and move the data generated across the organisation, centralise, cross reference and store it before processing.
Finally, we have the actual advances in AI, which are significant. Many applications are now impacting people in everyday fashion, for example self-driving cars and assistants like Siri and Alexa.
But despite these technological advances, deploying AI in the enterprise is harder than in a point application. Complexity arises from the multiple interfaces that must be used, which are accreted in enterprises as they incrementally evolve their IT systems overtime. Equally, new solutions, changes in business, and acquisition of other businesses have resulted in multiple sources of data and often varying CRM streams.
One response would be to undertake transformative investments to simplify and streamline IT, but this is risky if journeyed without the right transformation partner. Ultimately, IT continues to evolve towards complexity in a world where we can’t afford to rip it out and start again. This results in companies fearing delivering AI into the enterprise “ball of wool”.
So, what to do? My experience has led to three rules to follow when looking to implement AI in a large enterprise:
- No goldilocks: either invest at scale, so that you get sufficient capability to transform a core process with a very large reward, or focus on tactical peripheral processes with low execution risk. Avoid the middle ground. For complex processes you will have to build and sustain a team over a long period; you must anticipate really significant results to justify this, and the project must matter enough to executive management to be a persistent “must do” on their agenda.
- No heroes: this is a team game; build sustainable teams that are able to handle consultancy processes or can be embedded into business innovation over time. Innovations like AI shouldn’t be considered a standalone thing.
- Infrastructure for success: develop the hardware (both physical and virtual), social and knowledge infrastructures to support implementation. And this involves a clear adoption plan.
Hear more from Simon as he took to the stage at Innovation 2017. He’ll also be speaking at Codex’s Industries of the Future event, which BT are sponsoring, this September.