CIO OPINION change – with all the attending opportunities for fast growth – is guaranteed. Organisations must adapt quickly or be left in the wake of others.
Generative AI is only as powerful as the infrastructure that underpins it, so many organisations are either undertaking, or at least planning, large-scale improvements to their IT estates.
Generative AI means potentially vast amounts of data ingested by language models, large and small within your network, across different organisations and third party private and public organisations.
Hyperscalers help organisations to deal with the pace of change and the hard-to-predict impact through their flexible cloud computing resources, such as processing and storage. But the same flexibility is required for their connectivity.
To work effectively, and for the best results, AI relies on diverse data inputs and a variety of information that is both deep and wide. The source of these data inputs, and their size, are dictated by an organisation’ s AI strategy and use cases, plus its data governance framework.
LLMs, as well as small language models, smaller in scale and scope than their larger counterparts, use big data sets to create generalised learning structures, initially to train AI language models quickly; broader and more accurate insights are then derived from smaller sets.
All this needs AI-ready connectivity, but traditional network infrastructure has failed to keep pace and offers both poor performance and a poor user experience with long lead-times and lack of digital control. This is hindering organisations’ Generative AI adoption and transformation efforts.
They must strike the ideal balance between factors such as latency, bandwidth and cost. This means ensuring data is either processed efficiently locally, at the edge, or can travel freely without congestion to the large language model, LLM, often hosted in the cloud.
The telco edge can be considered the Goldilocks location, providing the ideal conditions to straddle edge and cloud. Providing seamless, dense interconnectivity at the telco edge, drives improved user experience for accessing digital apps and services.
Are traditional telco networks flexible and agile enough to cope?
To offer AI-interconnectivity, networks must adopt the same technical and commercial flexibility as hyperscalers in delivering computing power and storage.
Building AI capabilities is top technology priority for 11 % of corporations and 16 % of UK enterprises, according to recent research with ISG.
www. intelligentcio. com INTELLIGENTCIO MIDDLE EAST 41