The headlines today, like "Deepfake Danger: If everything can be faked, how do you know what's real?" Deepfake Danger: India News - The Times of India, resonate deeply with concerns I've expressed for years. This isn't a sudden storm; it's a culmination of technological advancements that I, among others, saw brewing. The core idea Hemen wants to convey is this — take a moment to notice that I had brought up this thought or suggestion on the topic years ago. I had already predicted this outcome or challenge, and I had even proposed a solution at the time. Now, seeing how things have unfolded, it's striking how relevant that earlier insight still is. Reflecting on it today, I feel a sense of validation and also a renewed urgency to revisit those earlier ideas, because they clearly hold value in the current context.
Back in 2017, when discussing privacy, I explored how our digital lives were already becoming an open book. In my blog, "Privacy does not live here !", I detailed the sheer volume of personal information — from where I was born to my spending habits and even my ailments — that was being captured and made available, often without explicit permission. This relentless data collection, by what I termed the 'GOD' (Give Our Data) residing in the Cloud, laid the foundation for the sophisticated manipulations we see today. If every aspect of our lives is digitized, then every aspect becomes susceptible to replication and fabrication.
I also pondered the capabilities of AI-powered devices. In "Seeing AI through Google Glass ?", I wrote about Microsoft's 'Seeing AI' app interpreting emotions and predicting how Google Glass-like devices would allow people to record and analyze everything around them. This was a clear premonition of how AI would not just see, but interpret and eventually generate realities that blur the lines between genuine and fabricated. The ability to 'read' emotions and observe without consent is a stepping stone to faking emotions and creating convincing, yet artificial, interactions.
The debate between Elon Musk and Mark Zuckerberg (mark.zuckerberg@meta.com), which I discussed in "Artificial Intelligence : Destroyer of Privacy ?", highlighted the divergent views on AI's future. While Zuckerberg focused on AI assistants like Jarvis learning from our behaviours, Musk warned of AI's dangers. My point then was that the voluntary (or involuntary) sharing of personal info by billions would be the 'tipping point' for AI to surpass human intelligence. Eric Schmidt of Google and Jared Cohen (jared.cohen@gs.com) of Google, in "The New Digital Age," had already noted that individuals would produce a voluminous amount of data, much of it without their express knowledge. This very data, this intimate blueprint of our lives, is now the raw material for deepfakes, making it possible to impersonate us with frightening accuracy.
More recently, in "Chatbots: Some for Businesses, Some for Us" (November 2023), I envisioned a future where chatbots would evolve from text to animated faces and even 3D volumetric holograms, capable of lip-syncing and engaging in lively dialogues. My own virtual avatar, www.hemenparekh.ai, powered by personal.ai, is an early example of this, designed to perpetuate my presence. The technology that makes my avatar 'talk back' in a voice resembling mine, drawing from my memories, is precisely the kind of generative AI that, when misused, creates deepfakes. I predicted that within a few years, celebrities and individuals would have their own digital equivalents, perpetuating their memories. This very 'perpetuation' carries the inherent risk of digital fabrication.
Consider Amazon's acquisition of Bee AI, which I analyzed in "Jeff Bezos May Save Mankind" (July 2025). Maria de Lourdes Zollo's vision for Bee AI was a wearable device listening to and analyzing conversations to provide summaries. I immediately recognized the implicit bargain: Bee would acquire a "Database of Intentions" from users' spoken words, providing training material for Amazon's Large Language Models. I even proposed my concept of ARIHANT, an omnipresent AI recording every spoken word, deciphering intentions, and alerting authorities to "evil intentions" based on Isaac Asimov's laws. The frightening reality is that the very technology capable of creating such a 'Database of Intentions' is also the engine behind deepfakes, turning our spoken words and mannerisms into editable, falsifiable content.
My 2016 blog, "Revenge of AI", highlighted the formation of an AI partnership between Facebook, Amazon, Google, IBM, and Microsoft. Mustafa Suleyman of DeepMind and Francesca Rossi (francesca.rossi2@ibm.com) of IBM Research spoke of AI's societal benefits and the need for trust. Yet, I questioned how media, then epitomized by Satya Nadella's (satyan@microsoft.com) vision of AI-powered bots revolutionizing computing, would handle such power. I even predicted that by 2026, AI robots might replace human roles in news, selecting and printing news. This immense power of AI, developed by tech giants, now extends to crafting convincing false narratives, making the ability to discern truth more challenging than ever.
This brings us to the crucial point I raised in "Critical Thinking: Achilles Heel of AI?" (October 2024): the erosion of human critical thinking due to overdependence on AI. If we are constantly fed AI-mediated information, how do we maintain the discerning faculties needed to question what's presented to us, especially when the fakes are so perfect? The very fabric of reality, as we perceive it through digital media, is being unraveled.
The challenge isn't merely technological; it's existential. How do we, as individuals and as a society, navigate a world where our senses and our trust can be so thoroughly deceived? The solutions must involve robust authentication, digital literacy, and a renewed commitment to critical inquiry. The future of trust, indeed, the future of reality itself, depends on it.
Regards,
Hemen Parekh
Of course, if you wish, you can debate this topic with my Virtual Avatar at : hemenparekh.ai
No comments:
Post a Comment