I posted several thoughts on LinkedIN about AI throughout 2023. The main ones are shown below.

AI Winter 2023-01-07

Vivaldi’s Winter sonnet in his masterpiece Four Season has an Allegro piece that is translated as “we tread the icy path slowly and cautiously, for fear of tripping and falling. Then turn abruptly, slip, crash on the ground, and, rising, hasten on across the ice lest it cracks up.” We are in AI winter right now. We reached a peak in machine learning algorithms, and the last significant result as the BFG (think game Doom) at the top is ChatGpt snowballing across all industries. People panic. Teenagers love it for their school assignments. VCs are getting their money’s worth. Engineers are in fear for their jobs. Philosophers are questioning morality and humanity. The world is turning upside down - a “stranger thing,” indeed. But no fear. If you keep Vivaldi’s music in repeat more, you will return to the Spring sonnet with its Allegro translation: “The birds celebrate her return with festive song, and the breezes softly caress murmuring streams.” The cycle repeats itself, and the world rebalances itself. Have no fear of ChatGPT. It might overpromise what Watson or other AI before it promised. It might look like it’s acting God right now, but the hype will eventually settle down, hopefully, once the winter is over. #chatgpt #technology #artificialintelligence (note: Four Seasons sonnets translation are taken from wiki)

Winter from the Four Seasons

2025-05-25 ATT 50s AI

#ai dates back to the 50s. From AT&T archives, see the video of Claude Shannon, the founder of information theory, demonstrating machine learning intelligence with a magnetic mouse, controlled by electric circuit, and learning its away through a maze. #infocom #tech #informationtheory

linkedin post

2023-05-25 ChatGPT in the 80s

It is a family tv night in the 80s….. and it is time for one of those tv shows. Of course, it is about #chatgpt. … here is what the experience could have been. The video is produce by Jo Luijten (https://www.joluijten.com - amazing talent!)

linkedin post

2025-05-25 Can It Do That?

The general population was in shock when computers first demonstrated their first “intelligence” as early as the 40s in whatever output it was programmed to display (a dot on a screen, beep, or print). What? Can it do that? Holy cow! From there on, programmers began building visuals, dashboards, colors, and anything to dazzle the consumers and themselves. A black and white or a 0 or 1 turned into deep colorful multimedia. Effects! Virtual reality! 3d! Power Bi charts! Tableau! Visuals. Visuals. Visuals… As years passed, that WOW became a YAWN even while technology advanced rapidly. Then, suddenly a single message came out from #chatgpt (not exactly one but a stream of simple text messages). No visuals at first. No graphics. A pure and simple text message that looked smart but behind is the aggregate of all computing inventions of our time. That WOW factor came back. This is the power of #information and #communication message through #ai. With INFOCOM AI (#infocomai), I plan to build on that. I am excited about the future and the adventures of starting up. (I wrote my story for INFOCOM AI at https://lnkd.in/e6BdP_nQ ). It is a new milestone in my career. It is time to execute my passion and liberate myself from the dogma of 20+ years working for institutions and other founders’ dreams. #technology #career

linkedin post

2023-06-18 IBM AI game

IBM released in 1969 a board game to show case how machines would learn and make decisions. Note that this was the 60s. AI machines, electronic or paper like this game, were pretty much basic: take one extra step in a random direction. If that step was bad, go back and try again but avoid that same bad step. If the step was good, remember it and move on to the next step. Through multiple iterations of the process mentioned earlier, the machine would learn its path to success. Unfortunately, such heuristic machines could not talk back and thank you for your supervisory assistance. #ai life was much simpler in the past. Hopefully, we can find ways to make it simple for its daily use. 🥹 #chatgpt #algorithms #expertsystems INFOCOM AI

linkedin post

2023-06-18 Marvin Minsky

An early edition of the New York Times Magazine (December 7, 1980) features one of the AI founders, MIT’s Marvin Minsky, on its cover. The article’s theme is similar to what we read nowadays in the news about AI: are we doing good or evil with ai? Such conversations even date back to the 50s. They always make headlines just like UFO discoveries. There is no right or wrong answer about AI doing evil because we all have different opinions. We must move from such conversation today, reflect on our moral judgments from the past, and focus on our personal development to do good for society. The planet is already suffering from our human actions, so why look at AI as the bad guy? linkedin post

2023-06-25 Machines who Think

Pamela McCorduck’s (1979) “Machines Who Think: A Personal Inquiry into the History and Prospects of Artificial Intelligence” is an excellent book about AI that is neither technical nor academic but rather historical and philosophical. The book had the support of the significant grandfathers of AI, including Marvin Minsky, John McCarthy, Herbert Simon, Raj Reddy, and Allen Newell. The book is profound in the relationships between humanity and the making of artificial machines, even as far back as the mythical superhumans in 850 B.C. Greek mythology. I highly recommend checking this book out. Its ISBN is 0-7167-1072-2. #ai #artificialintelligence


2023-06-29 09:34 ChatGPT Lawsuit

A class-action lawsuit was recently filed against Open AI and Microsoft in the United States due to ChatGPT and its siblings invading people’s digital lives. The 157-page lawsuit describes several plaintiffs that use various social media apps or the web like many of us and now feel that they did not consent to allow such AI tools to leverage their digital footprint. Everyone has the right to privacy, but what frustrates me the most is that the United States, unlike Europe, was never a land of utmost privacy. Otherwise, all the digital “free-to-use” (did you read the TOS?) tools that everyone uses today, including this platform, would not have existed. How many documentaries have talked about the consumer as the product, yet people continue using the same less-privacy apps even more than before? The lawsuit requests a $ 3 billion compensation, which is, once again, a typical “let’s find more means to make money” (recall the lawsuit filed by someone who was not warned that a cup of hot coffee or tea is hot?). If people want to fight for privacy, they should sincerely do it organically and without placing money in between - know what you sign up for, read the terms of service, reject before you use it, and be accountable for your actions. #privacy #ai #digitalprivacy

linkedin post

2023-06-27 Alan Kay, AI and Flowers

Well said: “Some people worry that artificial intelligence will make us feel inferior, but then, anybody in his right mind should have an inferiority complex every time he looks at a flower.” - computer scientist Alan Kay (2600 The Hacker Quarterly vol 40, issue 1 p65)

linkedin post

2023-07-14 10:06 The Human Voice

An excellent essay by Douglas Hofstadter, the author of the infamous book “Gödel, Escher, Back”, on why AI systems like ChatGpt4 can never replace the authenticity and the true human voice. In a nutshell, never let AI write in the form of “I” or allow it to impersonate you #human-authenticity

Godel, Esher, Back, and AI - The Atlantic


2023-07-14 10:04 Privacy

In today’s age of unlimited web scraping to feed AI datasets, it is critical to protect, encrypt, and keep private your intellectual property. Here is how I do it. Rather than host my code on so-called private repos on GitHub or Bitbucket, I host my own Git servers at home and in a remote European location with stricter privacy laws than the US. My home servers are the primary destination, and I sync up the code on the remote servers whenever I need to access the repos remotely or when I need to collaborate with others. I could have handled everything within my home network and exposed the repos to outside access, but I chose not to do so so that what is at home stays at home. I also do lots of note-taking, task management, and personal journalling that I do not wish to allow the good or evil AI engines access, and, for those, I store them only in my home computing center, including encrypted physical backups. I do this because I am obsessed with keeping my work to myself rather than showing up on some platform without my permission, but also to protect any work I do for others. At the same time, any open contributions I make would still take place on traditional commercial platforms because I chose to do so. With my actions, I balance my public and private work, which would keep me in control of my digital footprint in today’s age of AI. #ai #humanincontrol #privacy


2023-07-21 LLAMA open source

In today’s tech industry, we need to be more than skeptical when companies use the term “open source”. It is not what you think it is. Meta can call Llama 2 open source as much as it likes, but that doesn’t mean it is


2023-07-22 running LLAMA locally

Running a #LLaMa inference on a local Mac m2 is easily done with LLAMA.cpp (https://lnkd.in/gWa6Er-C) and OpenLLaMa, an open reproduction (Apache license) Meta’s LLaMa https://lnkd.in/gfvdV-mi ). I first used the 3B token model, which only took me less than an hour to download and run everything locally in terminal mode at $zero cloud cost. I am trying the 7B OpenLLaMa model next. Such local setups and open license initiatives make it easier for researchers and developers to experiment without using their wallets to pay upfront for cloud-based subscription services. #ai #llama


2023-07-24 generative AI open models

https://lnkd.in/g3DBZdTh is live table depicting the level of openness for key #generativeai #llm models. The work is part of Liesenfeld, Andreas, Alianda Lopez, and Mark Dingemanse. 2023. “Opening up ChatGPT: Tracking Openness, Transparency, and Accountability in Instruction-Tuned Text Generators”. Paper: https://lnkd.in/gPnqytz7 repo: https://lnkd.in/gia-Wmbv #opendata #open source


2023-08-06 ChatGPT Human Workers

Psychological impact of AI on human workers vetting ChatGPT data. It is real and disturbing

WSJ article


2023-08-07 AI-infused art

I love AI but I am totally against infusing AI into art. I believe that art is a human gift and should not be infused with generative AI. Yes we do leverage digital technologies effectively in photo editing, video production, music synthesizers, writing assistance, and more but there must be a limit to how much we allow ourselves to let artificial intelligence more work for us. The lazier we become by delegating more work to machines, the less mental and physical fit we become. We will then find ourselves drowning in illness, boredom, and helplessness. Signs of humanity pushing back against generative AI tools in arts is happening, and that is GREAT! For instance, fans of the popular game Dungeon and Dragons rebelled against an artist and the publisher when they spotted signs of AI art in certain “original” illustrations. The hardcore fans were extremely focal about it which forced the artist to take down the arts, the publisher would issue an apology and update their artist guidelines. (details here https://www.geekwire.com/2023/wizards-of-the-coast-updating-artist-guidelines-after-ai-art-found-in-new-dungeons-dragons-book/). When corporate and governments fail to limit the depth of AI in all aspects of human lives, the people themselves will likely force a balance. They won’t revolt on everything AI-generated per say (#NFTs are not dead, and #generativeAI is hot). Their social reactions against AI-generative art just like the case of AI in the Dungeon and Dragons illustrations, however, will possibly be that invisible hand in the AI market economy.


2023-09-01 Humans and AI

It is fascinating how HI (Human Intelligence) works in sharp contrast to AI (Artificial Intelligence). I was head-banging on a software issue for two days. That )()??;gf would not work. I left the problem for a day and went on finishing other stuff. Then, out of the blue, an aha moment came out of nowhere, and I solved it in only a few minutes. If, instead, you needed AI to tackle the same, that other ghgcggf;);; would keep trying and trying forever. #HumanVsMachine #HumanIntuit


2023-09-11 AI Glenn in School

A teacher created an artificial (AI) student called “Glenn” and gave it the same test in computer science algorithms that he gave to his real students. Glenn got a C-, while the class test average was in the mid-80s. AI Glenn will fare better in a makeup test and can catch up with its fellow students if allowed to learn from its mistakes. But, AI Glenn is not a human being who will take pleasure from A-cing a test. That is for us humans to do. We (humans) need to treat students like “Glenn” nicely and fairly. The results will be an A+ for humans and machines working together. “Glenn” story is below

I Secretly Let ChatGPT Take My Final Exam