recentpopularlog in

Copy this bookmark:





to read

bookmark detail

Everything still to play for with AI in its infancy
February 14, 2019 | Financial Times | by Richard Waters.

the future of AI in business up for grabs--this is a clearly a time for big bets.

Ginni Rometty,IBM CEO, describes Big Blue’s customers applications of powerful new tools, such as AI: “Random acts of digital”. They are taking a hit-and-miss approach to projects to extract business value out of their data. Customers tend to start with an isolated data set or use case — like streamlining interactions with a particular group of customers. They are not tied into a company’s deeper systems, data or workflow, limiting their impact. Andrew Moore, the new head of AI for Google’s cloud business, has a different way of describing it: “Artisanal AI”. It takes a lot of work to build AI systems that work well in particular situations. Expertise and experience to prepare a data set and “tune” the systems is vital, making the availability of specialised human brain power a key limiting factor.

The state of the art in how businesses are using artificial intelligence is just that: an art. The tools and techniques needed to build robust “production” systems for the new AI economy are still in development. To have a real effect at scale, a deeper level of standardisation and automation is needed. AI technology is at a rudimentary stage. Coming from completely different ends of the enterprise technology spectrum, the trajectories of Google and IBM highlight what is at stake — and the extent to which this field is still wide open.

Google comes from a world of “if you build it, they will come”. The rise of software as a service have brought a similar approach to business technology. However, beyond this “consumerisation” of IT, which has put easy-to-use tools into more workers’ hands, overhauling a company’s internal systems and processes takes a lot of heavy lifting. True enterprise software companies start from a different position. They try to develop a deep understanding of their customers’ problems and needs, then adapt their technology to make it useful.

IBM, by contrast, already knows a lot about its customers’ businesses, and has a huge services operation to handle complex IT implementations. It has also been working on this for a while. Its most notable attempt to push AI into the business mainstream is IBM Watson. Watson, however, turned out to be a great demonstration of a set of AI capabilities, rather than a coherent strategy for making AI usable.

IBM has been working hard recently to make up for lost time. Its latest adaptation of the technology, announced this week, is Watson Anywhere — a way to run its AI on the computing clouds of different companies such as Amazon, Microsoft and Google, meaning customers can apply it to their data wherever they are stored. 
IBM’s campaign to make itself more relevant to its customers in the cloud-first world that is emerging. Rather than compete head-on with the new super-clouds, IBM is hoping to become the digital Switzerland. 

This is a message that should resonate deeply. Big users of IT have always been wary of being locked into buying from dominant suppliers. Also, for many companies, Amazon and Google have come to look like potential competitors as they push out from the worlds of online shopping and advertising.....IBM faces searching questions about its ability to execute — as the hit-and-miss implementation of Watson demonstrates. Operating seamlessly in the new world of multi-clouds presents a deep engineering challenge.
artificial_intelligence  artisan_hobbies_&_crafts  automation  big_bets  brainpower  cloud_computing  contra-Amazon  cultural_change  data  digital_strategies  early-stage  economies_of_scale  Google  hit-and-miss  IBM  IBM_Watson  internal_systems  randomness  Richard_Waters  SaaS  standardization  value_extraction 
february 2019 by jerryking
view in context