Stéphane Dalbera’s Post

↫↫↫↫↫ ʟʟᴍs & ᴅᴇᴠ : ᴛʜᴇ sᴇʀɪᴏᴜs ᴘᴀʀᴛ ↬↬↬↬↬ The rapid ascent of large language models (LLMs) is a double-edged sword, fueling both innovation and an abundance of misinformation. It has sparked a wave of magical thinking, baseless assertions, sensationalized press releases treated as indisputable truths, and predictions lacking scientific grounding. This phenomenon manipulates FOMO (Fear of Missing Out), fosters irrational beliefs, and borders on a new form of mysticism—in essence, it perpetuates irrationality. In this context, serious works are rare and those of the CodeScene teams are to be praised. I'm sharing with you today this study conducted by Adam Tornhill, Markus Borg & Enys Mones 🫡 𝘛𝘩𝘪𝘴 𝘳𝘦𝘱𝘰𝘳𝘵 𝘪𝘴 𝘵𝘩𝘦 𝘤𝘰𝘯𝘤𝘭𝘶𝘴𝘪𝘰𝘯 𝘰𝘧 𝘢 𝘣𝘦𝘯𝘤𝘩𝘮𝘢𝘳𝘬 𝘴𝘵𝘶𝘥𝘺 𝘰𝘧 𝘵𝘩𝘦 𝘮𝘰𝘴𝘵 𝘱𝘰𝘱𝘶𝘭𝘢𝘳 𝘓𝘢𝘳𝘨𝘦 𝘓𝘢𝘯𝘨𝘶𝘢𝘨𝘦 𝘔𝘰𝘥𝘦𝘭𝘴 (𝘓𝘓𝘔𝘴) 𝘢𝘯𝘥 𝘵𝘩𝘦𝘪𝘳 𝘢𝘣𝘪𝘭𝘪𝘵𝘺 𝘵𝘰 𝘨𝘦𝘯𝘦𝘳𝘢𝘵𝘦 𝘤𝘰𝘥𝘦 𝘧𝘰𝘳 𝘳𝘦𝘧𝘢𝘤𝘵𝘰𝘳𝘪𝘯𝘨 𝘵𝘢𝘴𝘬𝘴. 𝘞𝘦 𝘢𝘪𝘮 𝘵𝘰 𝘪𝘭𝘭𝘶𝘴𝘵𝘳𝘢𝘵𝘦 𝘵𝘩𝘦 𝘤𝘶𝘳𝘳𝘦𝘯𝘵 𝘴𝘵𝘢𝘯𝘥𝘢𝘳𝘥𝘴 𝘢𝘯𝘥 𝘭𝘪𝘮𝘪𝘵𝘢𝘵𝘪𝘰𝘯𝘴, 𝘢𝘯𝘥 𝘴𝘦𝘦𝘬 𝘵𝘰 𝘴𝘩𝘰𝘸 𝘯𝘦𝘸 𝘮𝘦𝘵𝘩𝘰𝘥𝘰𝘭𝘰𝘨𝘪𝘦𝘴 𝘸𝘪𝘵𝘩 𝘩𝘪𝘨𝘩𝘦𝘳 𝘤𝘰𝘯𝘧𝘪𝘥𝘦𝘯𝘤𝘦 𝘳𝘦𝘴𝘶𝘭𝘵𝘴. #LLMs #IA #Dev #Refactoring

¸,ø¤º°`°º¤ø,¸ 𝐁𝐞𝐜𝐚𝐮𝐬𝐞 𝐢𝐭’𝐬 𝐅𝐫𝐢𝐝𝐚𝐲 : 𝐃𝐨𝐧𝐚𝐥𝐝 𝐊𝐧𝐮𝐭𝐡 ¸,ø¤º°`°º¤ø,¸ To wrap up a week dedicated to celebrating the gold-standard geek humor of the great 𝐃𝐨𝐧𝐚𝐥𝐝 𝐊𝐧𝐮𝐭𝐡, I'm continuing with my #BecauseItsFriday by republishing this. ---- 𝐊𝐧𝐮𝐭𝐡's first scientific publication was in one of the most prestigious and selective peer-reviewed journals: Mad 🫡 𝘒𝘯𝘶𝘵𝘩 𝘱𝘶𝘣𝘭𝘪𝘴𝘩𝘦𝘥 𝘩𝘪𝘴 𝘧𝘪𝘳𝘴𝘵 “𝘴𝘤𝘪𝘦𝘯𝘵𝘪𝘧𝘪𝘤” 𝘢𝘳𝘵𝘪𝘤𝘭𝘦 𝘪𝘯 𝘢 𝘴𝘤𝘩𝘰𝘰𝘭 𝘮𝘢𝘨𝘢𝘻𝘪𝘯𝘦 𝘪𝘯 1957 𝘶𝘯𝘥𝘦𝘳 𝘵𝘩𝘦 𝘵𝘪𝘵𝘭𝘦 “𝘛𝘩𝘦 𝘗𝘰𝘵𝘳𝘻𝘦𝘣𝘪𝘦 𝘚𝘺𝘴𝘵𝘦𝘮 𝘰𝘧 𝘞𝘦𝘪𝘨𝘩𝘵𝘴 𝘢𝘯𝘥 𝘔𝘦𝘢𝘴𝘶𝘳𝘦𝘴”. 𝘐𝘯 𝘪𝘵, 𝘩𝘦 𝘥𝘦𝘧𝘪𝘯𝘦𝘥 𝘵𝘩𝘦 𝘧𝘶𝘯𝘥𝘢𝘮𝘦𝘯𝘵𝘢𝘭 𝘶𝘯𝘪𝘵 𝘰𝘧 𝘭𝘦𝘯𝘨𝘵𝘩 𝘢𝘴 𝘵𝘩𝘦 𝘵𝘩𝘪𝘤𝘬𝘯𝘦𝘴𝘴 𝘰𝘧 𝘔𝘢𝘥 𝘕𝘰. 26, 𝘢𝘯𝘥 𝘯𝘢𝘮𝘦𝘥 𝘵𝘩𝘦 𝘧𝘶𝘯𝘥𝘢𝘮𝘦𝘯𝘵𝘢𝘭 𝘶𝘯𝘪𝘵 𝘰𝘧 𝘧𝘰𝘳𝘤𝘦 “𝘸𝘩𝘢𝘵𝘮𝘦𝘸𝘰𝘳𝘳𝘺”. 𝘔𝘢𝘥 𝘱𝘶𝘣𝘭𝘪𝘴𝘩𝘦𝘥 𝘵𝘩𝘦 𝘢𝘳𝘵𝘪𝘤𝘭𝘦 𝘪𝘯 𝘪𝘴𝘴𝘶𝘦 𝘕𝘰. 33 (𝘑𝘶𝘯𝘦 1957). ᴛʜᴇ ɢᴏᴀᴛ🖖 #geekhumor #BecauseItsFriday

↫↫↫↫↫ ᴋɴᴜᴛʜ's ʙᴏᴜɴᴛʏ ʜᴜɴᴛᴇʀ ↬↬↬↬↬ This morning, François 🐙 Pacull and I discussed the various awards and honors available to computer science researchers. There is a very prestigious but little-known award that recipients proudly frame and display on their office walls : The 𝐊𝐧𝐮𝐭𝐡 reward checks 🤑 𝐊𝐧𝐮𝐭𝐡 reward checks are unique checks or certificates given by computer scientist 𝐃𝐨𝐧𝐚𝐥𝐝 𝐊𝐧𝐮𝐭𝐡 to individuals who identify technical, typographical, or historical errors, or offer significant suggestions for his publications. But 𝐊𝐧𝐮𝐭𝐡 being the gold standard of geek humor : In the preface of each of his books and on his website, 𝐃𝐨𝐧𝐚𝐥𝐝 𝐊𝐧𝐮𝐭𝐡 offers a reward of $2.56 to the first person who finds an error in his published works, whether it be technical, typographical, or historical. 𝐊𝐧𝐮𝐭𝐡 explains that $2.56, or 256 cents, is equivalent to one hexadecimal dollar. Initially, 𝐃𝐨𝐧𝐚𝐥𝐝 𝐊𝐧𝐮𝐭𝐡 sent real, negotiable checks to recipients who found errors in his works, but he stopped in October 2008 due to check fraud. Instead, he established the fictional "Bank of San Serriffe" and now sends "hexadecimal certificates." By October 2001, Knuth had written over 2,000 checks with an average value of over $8 each, and by March 2005, the total value exceeded $20,000. Most recipients framed these checks as trophies rather than cashing them.

╰☆☆ ᴡʜᴏ sᴀʏs sᴄɪᴇɴᴛɪsᴛs ᴀʀᴇɴ'ᴛ ɪɴᴛᴇʀᴇsᴛᴇᴅ ɪɴ ᴍᴀʀᴋᴇᴛɪɴɢ? ☆☆╮

𝐆𝐞𝐫𝐚𝐥𝐝 𝐖𝐞𝐢𝐧𝐛𝐞𝐫𝐠 (1933-2018) was an influential American computer scientist, author, and teacher known for his work in software development and systems thinking. He earned his PhD in Communication Sciences from the University of Michigan in 1963. Weinberg worked on Project Mercury at IBM and later became a consultant and prolific author, publishing over 40 books and 400 articles. His notable works include "The Psychology of Computer Programming" and "Introduction to General Systems Thinking". He was also a visiting professor and received several awards for his contributions to software engineering. I'm sharing this humourous interview response with you 👀

Never overestimate the human capacity to conceive and implement complex long-term strategies. Never underestimate the human capacity to commit foolish acts.

🧡❤️️ 𝓝𝓮𝓧𝓣 𝓶𝔂 𝓛𝓸𝓿𝓮 ❤️️🧡 I was fortunate to use Unix workstations as a teenager, but some machines remain particularly dear to me, like the 𝐍𝐞𝐗𝐓 𝐜𝐨𝐦𝐩𝐮𝐭𝐞𝐫𝐬. My fondness for these machines clearly comes from their exceptional design (crafted by Hartmut Esslinger for the hardware and 𝐏𝐚𝐮𝐥 𝐑𝐚𝐧𝐝 for the logo), their technical capabilities including the integration of a 𝐌𝐨𝐭𝐨𝐫𝐨𝐥𝐚 𝟓𝟔𝟎𝟎𝟏 𝐃𝐒𝐏, and a remarkable operating system largely developed by two Frenchmen, Bertrand Serlet and the late Jean-Marie Hullot, who would later become a friend.🫡 My attachment to these machines is also tied to their pivotal role in my first entrepreneurial venture. Alongside developing documentary research software on the 𝐈𝐁𝐌 𝐑𝐒/𝟔𝟎𝟎𝟎, we leveraged the 𝐍𝐞𝐗𝐓's 𝐃𝐒𝐏 to innovate new types of musical applications, similar to what IRCAM would pursue. But my most cherished memory of 𝐍𝐞𝐗𝐓 is undoubtedly the joy of programming it provided, especially with Jean-Marie's interface-builder, which was revolutionary at the time. During this period, I also had the opportunity to meet several notable figures, including 𝐒𝐭𝐞𝐯𝐞 𝐉𝐨𝐛𝐬 and, of course, Jean-Marie. Although the 𝐍𝐞𝐗𝐓 venture didn't achieve the success initially hoped for, it laid the groundwork for Apple's resurgence following the acquisition. Today, 𝐍𝐞𝐗𝐓 's legacy can still be seen in all Apple products. Here is 𝐁𝐲𝐭𝐞's article on the launch.

↫↫↫↫↫ ʟᴇᴛ's ᴘᴜᴛ ᴛʜᴇ ᴅɪsᴛᴀɴᴄᴇs ɪɴᴛᴏ ᴘᴇʀsᴘᴇᴄᴛɪᴠᴇ. ↬↬↬↬↬ Space distances are almost unimaginable, so let's take a look at our "progress". The human-made object that has traveled the furthest from Earth is the space probe 𝐕𝐨𝐲𝐚𝐠𝐞𝐫 𝐈, which, since its launch on September 5, 1977 (46 years ago), has reached a distance of 𝟐𝟐.𝟑𝟑 𝐥𝐢𝐠𝐡𝐭-𝐡𝐨𝐮𝐫𝐬 from Earth. For manned flights, the record belongs to the 𝐀𝐩𝐨𝐥𝐥𝐨 𝟏𝟑 crew. Although they didn't land on the moon, they hold the record for the farthest distance from Earth: 𝟏.𝟑𝟑 𝐥𝐢𝐠𝐡𝐭-𝐬𝐞𝐜𝐨𝐧𝐝𝐬 on April 15, 1970. As a reminder, the nearest star is 𝐏𝐫𝐨𝐱𝐢𝐦𝐚 𝐂𝐞𝐧𝐭𝐚𝐮𝐫𝐢 at 𝟒.𝟐𝟔 𝐥𝐢𝐠𝐡𝐭-𝐲𝐞𝐚𝐫𝐬 away. #Space #Astronomy #Science #Astronautics

╰☆☆ ᴀɴᴏᴛʜᴇʀ ᴘᴏᴡᴇʀғᴜʟ ғᴇᴀᴛᴜʀᴇ ᴏғ ʙɪᴛᴡɪsᴇ ᴏᴘᴇʀᴀᴛᴏʀs ☆☆╮ I regularly emphasize the importance of understanding the internal representation of the data you are manipulating, as well as the power of bitwise operators. ---- A few days ago, I came across a post on LeetCode where the challenge was to determine whether an integer is a power of two. The proposed solutions ranged from using the mathematical library to calculate logarithms to relatively complex algorithms. However, there's a much simpler and efficient solution with 𝙊(1) complexity 𝐧 & (𝐧 - 𝟏) == 𝟎 𝐚𝐧𝐝 𝐧 > 𝟎 That all folks 🤓 The binary representation of a power of 2 contains only a single bit set to 1. The binary representation of a Mersenne number (2^n - 1) has all bits to the right of the bit representing 2^n set to 1, with all other bits set to 0. #journey2theroots