A chance to do things better: AI vs. Internet

An AI generated image I designed to represent an unfettered internet and a still youthful AI

 

Being a late Gen-Xer, I was an adolescent when the internet fundamentally disrupted most of the world's access to information. Its mainstream deployment gathered steam in the 1990s, starting with text based interactions between humans, the publishing of HTML formatted content, and gave rise to the usefulness of not only portable telephones, but devices with large screens, to wit, smart phones (Japan had led this technological invention, for reasons such as but not limited to having double-byte characters with bifold designs of phones with colour displays and beautiful ringtones. But this was around 2000-2005: then the iPhone arrived on the scene). 

What I did not see these past 30-ish years, however, was governments and public services thinking through how to reach their voting constituents or their dependents. Nor, during my many years strategizing about search engine optimization, social media marketing, and web site enhancement topics, did I discern any overlay of ethical frameworks on how the internet itself should be leveraged to serve the population. GDPR compliance and other data privacy related notions have been challenging to implement precisely because, like many OWASP vulnerabilities, they have been added post hoc to an existing ecosystem. 

Instead, we with scant exceptions have been using Big Tech owned Social Media channels to disseminate news, facilitate schoolwork, and otherwise only designed custom applications when data deemed more sensitive (e.g. taxation related PII/SPI) came into play. The sheer convenience (and possibly an aversion to "reinventing the wheel"?) and cost factors appear to be mainly responsible for our situation. After widespread awareness of "political technology" and the takeover of popular channels by billionaires, next gen platforms such as Bluesky and Substack have arisen.

And now, a huge swathe of humans have access to generative AI solutions, much of which has been trained to date on internet content which reflects biases, myths, prejudices, and blatant misinformation which humans have been creating for decades. This means that "we" (human enabled AI) can publish even more "information" online which is just as likely built upon a foundation of contradictory or fictitious . We find ourselves needing to take a step back to examine how we can quality control the inputs we feed AI.

AI deployment in specific contests already seems to be having some false starts - the hype and promises are over inflated as with anything novel, and I am sincerely hoping that societies will choose to curb real life consequences humanely and sensibly even as we increasingly rely on AI and come to terms with how it, as any powerful tool, can be harnessed for good and monitored for misuse/wrongdoing.

There have, fortunately, long existed more authoritative, powerful voices than mine advocating for ethical AI, curating training data stringently, and wider adoption of medium or smaller language models for deeply specialized and accurate insights. And my path to becoming an expert myself in any specific aspect of this will be a long one despite my grounding in natural language processing and hands on work in machine learning. Nevertheless, I am one more concerned citizen, insisting that all technology benefits from being steered towards a balanced and thoughtful approach between innovation and societally responsible implementation. Let us use the internet, which connects us all, for this ethical and valuable aim, so that we handle AI better than we did with the internet was in its infancy.