AI algorithms
Signal #3 - Weekly insights to rise above the sea of noise.
đ Trend - AI algorithms evolution and impacts
Artificial Intelligence (AI) algorithms are excelling in areas we once thought to be uniquely human, including prediction, recommendation, and now even creative tasks like drawing. This evolution will unlock substantial economic value by increasing the amount of intelligence we can allocate to many applications, such as detecting cancer, optimising traffic, matchmaking âŠ
However, as we edge closer to the almost divine capability of creating intelligent entities, I am reminded of a quote by the Renaissance writer François Rabelais: âScience without conscience is but the ruin of the soul.â With this in mind, this week we will focus on the history of applying AI in social media, one of the earliest, most impactful and well funded applications of AI.
A short history of AI in social media
With the introduction of the "Customers Who Bought This Item Also Bought" feature in 1990, Amazon pioneered the recommendation system. Shortly thereafter, social media platforms like Facebook adopted the same AI algorithm, known as collaborative filtering, to enhance their services.
This marked a significant shift from the traditional network of relationships, characterised by the âSubscribe / followâ method of organising content, to an AI-enhanced network model that more actively determined what content you see by ranking it. This innovation enabled companies to increase user engagement on their platforms, but it also exacerbated issues such as echo chambers and the spread of misinformation.
TikTok entirely abandonned the conventional âSubscribe/Followâ model by introducing the âFor Youâ page, significantly increasing user interaction through deep learning algorithms. This demonstrated the potential of AI to predict user interests more accurately than users themselves, prompting platforms like YouTube and Instagram to implement similar features with Shorts and Reels, respectively.
This deep-learning approach not only increased user engagement further but also incentivised a trend-centric content ecosystem. This posed a threat to the creator economy by shifting the balance of power to the platforms rather than the creators and discouraged originality. However, itâs also a huge advancement in the organization of information and software personalization.
We are now entering into a new era of AI integration in social media with the emergence of generative AI. This technology enables âfakeâ AI influencers, (ie algorithms as our role models), and is becoming an integral part of creator workflows, blending AI with human creativity to produce content.
To summarise, this history of AI implementation in social media showcases:
Alignment issues: AI can amplify technology's drawbacks, such as increased screen time and echo chambers, if user interest is not at the center.
Disruption potential: AI can rapidly and unpredictably disrupt industries, yet also boost productivity and personalise software. This also showcases an application where AI didnât replace existing jobs but created new value (no one was organising content before).
Black Box effect: With more reliance on AI, which could arguably be viewed as csv (ie Excel) files with billions of parameters, we might go into a world where the link between input and output is more complex (eg. creators unsure what videos will be put forward by algorithms). A scenario where AI systems optimize both content creation and curation, may lead to questionable outcomes. (other example: AI to optimise resumes and AI to screen resumes).
As these algorithms get more and more powerful, I believe the central challenge, and our responsibility, in AI development lies in ethical principles to ensure technology enhances human life without compromising our values or societal wellbeing. Particularly important might be how to make sure private companies, ever more powerful, prioritise enhancing long term human benefits over short term profits.
Disclaimer: AI was used as a brainstorming assistant, co-writer, and proofreader in the creation of this article. :)
đĄ Insight - Writing to Learn
Building on the popular saying "If you can't explain it simply, you don't understand it well enough.", structuring your thoughts for others to read will help you learn and clarify your ideas.
đ Links - Check these out
Dopamine Nation - We explored some of the downsides of AI in social media, one of which is creating addictive behaviors. This book by psychiatrist Dr. Anna Lembke, delves further into some of the impacts and mitigation strategies in a dopamine rich world.
Writing to Learn - Explore the concept of writing to learn further, with this book by William Zinsser.