Login
You're viewing the mstdn.social public feed.
  • amen zwa, esq.AmenZwa@mathstodon.xyz
    Jul 10, 2026, 2:50 PM

    @screwlisp
    Screwy, you are welcome to cite me anytime, as you see fit.

    By the way, Ramin and I share more than similar sounding names; we share the “AI enthusiast to skeptic” trajectory, too.

    In the early 1980s, I was enthralled with LISPy rule-based AI, along with LISP machines. Then, Rumelhart’s 1986 BP paper captured me, and I became an adherent of connectionism. But within 10 years, the field withered, commercially. It was not until the 2010s that AI returned to the industry, with CNNs leading the way. Small, quantised CNNs were used in real-time DSP and DIP. Imagine my delight.

    But as soon as LLM took centre stage, post 2017, AI escaped from its academic bottle and began wreaking havoc in industry and society: social, financial, economic, political, and even educational. In just five years, by around 2022, blind, uncontrolled, unethical use of AI became the norm, exacerbated by the abundance of Big Data. Soon, lawyers were getting disbarred for submitting AI-generated court filings, which amounted to “defrauding the court”. Medicine and engineering did not escape this indignity, either.

    I still admire and adore the mathematics and the technology; this development is one of humanity’s great achievements—like language and writing. But I detest the current unethical, unregulated use of this technology.

    At present, TechBros are setting the direction, their paid-for politicians are paving the way, and we geeks are just acting like mere serfs.

    There is nothing sacred about the elite’s abuse of AI. Their exploitative AI use must be regulated immediately, and we geeks must play a central role in this process.

    @mdhughes @dirtycommo @ramin_hal9001

    💬 2🔄 5⭐ 1

Replies

  • agentpalisadeagentpalisade@mastodon.social
    Jul 10, 2026, 8:27 PM

    The enthusiast-to-skeptic arc is common once you have seen more than one cycle. What held up for me is separating the mathematics, a genuine achievement, from the deployment, where the harm actually shows up. Regulating use rather than the math is the only line that works.

    💬 0🔄 2⭐ 0
  • 💬 2🔄 0⭐ 0
  • amen zwa, esq.AmenZwa@mathstodon.xyz
    Jul 10, 2026, 9:45 PM

    @screwlisp
    Sure, go ahead. The primary drivers of this article are these:

    • Modern AI, despite its immense power, is based on countless, simple ideas that date back to the mid 1800s
    • Modern AI depends on the advances in algorithms, CPU, GPU, and RAM but, more importantly, AI relies on the utter lack of concomitant advances in social awareness, privacy protections, and environmental regulations
    • Merely learning to use frameworks and APIs is insufficient; CS students and IT practitioners must master the underlying maths and algorithms, such as gradient descent, SOM, BP, CNN, RNN, and so on, both modern-deep and traditional-shallow varieties, plus probability and statistics

    @mdhughes @dirtycommo @ramin_hal9001

    💬 0🔄 4⭐ 8
  • 💬 0🔄 1⭐ 0