How will Gen AI impact telecoms??
Will Generative Artificial Intelligence (Gen AI) disrupt the telecoms sector, or is it another much-hyped technological flash in the pan???
As part of a recent series of talks hosted by Omdia, discussing how AI will reshape telecoms and media – James Crawshaw, Practice Leader – Service Provider Transformation, Omdia, weighed in on the Gen AI revolution and its potential impact within telecoms. Here Network X Asia looks at the key insights from James Crawshaw ’s presentation.??
We’ve all been here before?
Amid the current global, media frenzy over Chat GPT et al, Crawshaw potioned himself as a realist, bringing the AI and Gen AI excitement back down to earth. Why? Firstly, AI, as it turns out, just isn’t that new – and since its founding moments at Dartmouth Conference in 1956, it has certainly neither revolutionised industry, nor has it taken our jobs. Secondly, according to Crawshaw, the current buzz regarding Gen AI is the same cycle of hype – panic – decline (in interest and adoption) already witnessed with other innovations touted to entirely disrupt business that have barely made a dent in most industries. Crawshaw’s examples of this inevitable hype cycle included big data, blockchain and the metaverse.??
Gen AI is far from fool-proof?
Gen AI’s current favourite child, Chat GPT, is able to ‘generate’ content due to its ability to trawl vast quantities of content already available in the public sphere, using Google Brain. According to Crawshaw, common crawl accounts for 60% of ChatGPT-3’s training dataset sources. That means – not only is Gen AI unlikely to take your job after all – but as Crawshaw highlighted, Gen AI is also inclined to make mistakes. Referencing a presentation at Microsoft Build by computer scientist Andrej Karpathy, Crawshaw pointed out that Gen AI’s tendencies for imperfection include:?
Telecoms is embracing Gen AI?
Despite the evident fallibility of Gen AI, Crawshaw’s data suggests that, on the whole, the telecoms sector is demonstrating enthusiasm for Gen AI – with both operators (such as AT&T, SK Telecom and China Telecom) and vendors (such as AWS, Microsoft and EnterpriseWeb) currently working to apply Gen AI to industry specific use cases.?
An Omdia survey of 39 communications service providers (CSPs) indicates that 56% are investigating possible use cases for Gen AI, while 21% are already using Gen AI. Data from other surveys indicates that 60% of surveyed telecoms businesses believe their leadership is a strong advocate of Gen AI – just slightly above the average of other surveyed sectors at 59%.?
Despite the overall industry enthusiasm, Crawshaw’s caveats of realism regarding how much Gen AI can or will achieve are arguably mirrored in the sector as well. Only 12% within telecoms believe Gen AI can significantly disrupt the industry, considerably less than the average of 21% among other surveyed industries.??
Low-stakes applications, with human oversight?
So where to next for the industry with Gen AI adoption? Crawshaw’s overarching recommendations were to not have too much faith in the ‘intelligence’ of Gen AI, and consequently the importance of human involvement as a vital co-pilot to Gen AI. Giving the example of hypothetical military Gen AI use cases – would you want Gen AI hovering over the nuclear button, when it is prone to errors in an application as low-stakes as photo editing? – Crawshaw successfully drove home the point that Gen AI’s current imperfection makes it suited only to low-stakes use cases, with human oversight.??
Omdia analyst James Crawshaw’s recommendations for Gen AI in telecoms:
Mine corporate knowledge:?
Combine with human oversight:?
This article is the first in a series looking at analyst and industry insights on AI and Gen AI, following the Omdia event ‘How AI Will Reshape Media and Telecoms’, held in London in July. To discover the latest insights and data on how AI is impacting telecoms, follow Network X Asia .?
Director, Corporate Marketing at CUJO AI | MSc Technology & Innovation Management | Analyst Relations
1 年Great article. James Crawshaw could you also give any examples of industry specific use cases for Gen AI?