As AI’s nascent development indicates, significant human intervention remains necessary to guide its progress. As Group Management Core’s founder, even the mere intention to choose between large language models is “inherently complex and daunting for most individuals.” Once you’ve selected what you deem the “right” model (whatever that means), you still need to fine-tune it to produce relevant outcomes – let alone achieving consistent results, which is not a feature currently inherent in LLMs.
After requesting RedMonk co-founder James Governor whether AI/general AI had misplaced its sheen, he replied with an unequivocal “No”. It’s possible that at present we’re experiencing the trough of disillusionment – a phrase I borrowed rather than his own – yet this is solely because new technologies invariably follow a similar trajectory: from indifference to adoration, then scorn and finally widespread acceptance. As some software developers rapidly advance to the final stage, others will face additional hurdles.
Finally constant
It’s become increasingly evident that the company needs to hit its stride effectively to achieve success. It’s said that merely fidgeting with something like Midjourney can conjure an image before you find it, yet “most AI art tends towards kitsch.” Does this stem from computers’ lack of understanding of what constitutes good art? As Grady Booch, a renowned AI skeptic, astutely observes: AI does not genuinely have the capability to cause or suppose, for both claims are unfounded and lacking empirical support. While comparing human thought processes with large language models (LLMs), it is essential to acknowledge that human consideration and comprehension are fundamentally distinct from statistical calculations, and reducing them to such processes would be a misinterpretation of humanity’s cognitive essence.
