[GUEST ACCESS MODE: Data is scrambled or limited to provide examples. Make requests using your API key to unlock full data. Check https://lunarcrush.ai/auth for authentication information.] #  @ChombaBupe Chomba Bupe Chomba Bupe posts on X about ai, artificial intelligence, tokenization, future the most. They currently have XXXXXX followers and 2137 posts still getting attention that total XXXXXXX engagements in the last XX hours. ### Engagements: XXXXXXX [#](/creator/twitter::1248939194/interactions)  - X Week XXXXXXX -XX% - X Month XXXXXXX +101% - X Months XXXXXXXXX -XX% - X Year XXXXXXXXX -XX% ### Mentions: XX [#](/creator/twitter::1248939194/posts_active)  - X Week XX -XX% - X Month XXX +127% - X Months XXX +4.30% - X Year XXX -XX% ### Followers: XXXXXX [#](/creator/twitter::1248939194/followers)  - X Week XXXXXX +0.03% - X Month XXXXXX +0.58% - X Months XXXXXX +7.20% - X Year XXXXXX +45% ### CreatorRank: XXXXXX [#](/creator/twitter::1248939194/influencer_rank)  ### Social Influence **Social category influence** [finance](/list/finance) [celebrities](/list/celebrities) [stocks](/list/stocks) [technology brands](/list/technology-brands) [countries](/list/countries) **Social topic influence** [ai](/topic/ai), [artificial intelligence](/topic/artificial-intelligence) #613, [tokenization](/topic/tokenization) #20, [future](/topic/future), [generative](/topic/generative), [elon musk](/topic/elon-musk) #1676, [6969](/topic/6969), [toy](/topic/toy), [agi](/topic/agi), [settings](/topic/settings) **Top assets mentioned** [ON Semiconductor, Corp. (ON)](/topic/on-semiconductor) ### Top Social Posts Top posts by engagements in the last XX hours "AI is usually trained and tested in toy problems. That's what benchmarks are toy representations of a tiny part of reality. The true test of intelligence is in the ability to get desirable outcomes in reality while facing ambiguity uncertainty and constant change" [X Link](https://x.com/ChombaBupe/status/1999279742844043737) 2025-12-12T00:46Z 11.8K followers, XXX engagements "This is an example of when you "teach" machine learning (ML) how to count oranges they learn to count oranges not the abstract concept of counting" [X Link](https://x.com/ChombaBupe/status/1999292629432041503) 2025-12-12T01:38Z 11.8K followers, 341.6K engagements "In this case maybe the model treats R and r differently but why The assumption should be that these are the same thing that is what abstraction is. Writing run and Run changes nothing" [X Link](https://x.com/ChombaBupe/status/1999293210368311304) 2025-12-12T01:40Z 11.8K followers, 21.3K engagements "Childrens aren't raised in highly controlled environments the way artificial intelligence (AI) is trained. There is a huge difference between winning a chess game and actually getting desirable results in noisy ambiguous environments like learning to cook" [X Link](https://x.com/ChombaBupe/status/1999279052818116908) 2025-12-12T00:44Z 11.8K followers, XXX engagements "You could also attribute this error to tokenization. But tokenization is a key component of these systems. What other problems is tokenization introducing in the models" [X Link](https://x.com/ChombaBupe/status/1999293572903022973) 2025-12-12T01:41Z 11.8K followers, 19.7K engagements "". these systems show some genuine fluid intelligence on simple tasks." Eventually artificial intelligence (AI) systems in controlled settings always beat humans ie chess Go. The challenges are always about extending such advances to handle real world complexity & ambiguity" [X Link](https://x.com/ChombaBupe/status/1999265988613931243) 2025-12-11T23:52Z 11.8K followers, 1971 engagements "I also don't publish because this is a one shot lifetime project. Its either I achieve the goal or I don't. There is no point for me to write intermediate research paper for what To appease who I want to hit the final mark" [X Link](https://x.com/ChombaBupe/status/1999681179017261118) 2025-12-13T03:22Z 11.8K followers, XXX engagements "Corporations/startups have vested interest to exaggerate the capabilities & future potential of their artificial intelligence (AI) products/services in order to sell them. There was no one selling fire so fire was not overhyped. Why is it so hard to understand" [X Link](https://x.com/ChombaBupe/status/1997177259036492164) 2025-12-06T05:32Z 11.8K followers, 7499 engagements "Making exaggerated claims like: Devin the first AI software engineer that is now no where to be seen superintelligence more intelligent than all X billion humans next year blah blah. Language models (LM) understand they are conscious. LMs will cause human extinction etc" [X Link](https://x.com/ChombaBupe/status/1997177747198660700) 2025-12-06T05:34Z 11.8K followers, XXX engagements "P = X the event is impossible observing it grants you infinite information like jaw dropping levels of events were the jaw drops to infinity ie escaping the black holes event horizon and telling the tale" [X Link](https://x.com/ChombaBupe/status/1998473092071567769) 2025-12-09T19:21Z 11.8K followers, XXX engagements "That being said let's go back to generative artificial intelligence (GenAI) and see why the information gain by observing their outputs is low. Let's start from first principles. A GenAI model G tries to generate outputs Y that follow a real distribution P" [X Link](https://x.com/ChombaBupe/status/1998473752800276955) 2025-12-09T19:24Z 11.8K followers, XXX engagements "If generative models didn't dependent on lots of highest quality data & still worked exactly like what they do now even with hallucinations today would be completely different. But in reality you have systems trained on literally "everything" yet still makes silly mistakes" [X Link](https://x.com/ChombaBupe/status/1998488532516835394) 2025-12-09T20:22Z 11.8K followers, 1103 engagements "I am a chill guy who just want a more capable efficient and ethical artificial intelligence (AI) system that would truly change the future. I don't hate AI I am simply want the AI of the future to be better than whatever it is today. There are better ways to do AI than this" [X Link](https://x.com/ChombaBupe/status/1998489222991458443) 2025-12-09T20:25Z 11.8K followers, XXX engagements "What this fails to touch on is that not all forms of remixing are the same. There are levels to these things randomingly shuffling words is essentially also writing but is it better than a deeply thought out written masterpiece Absolutely not" [X Link](https://x.com/ChombaBupe/status/1998644185176617414) 2025-12-10T06:41Z 11.8K followers, 2087 engagements "@AndyXAndersen Hardware is a limiting factor as well because single neurons do sophisticated things which would demand lots of energy if those functionalities were to be replicated on semiconductor digital computing systems" [X Link](https://x.com/ChombaBupe/status/1998947792442765614) 2025-12-11T02:47Z 11.8K followers, XX engagements "Normally when you isolate a particular problem well enough it can be solved better with automation since the problem is clearly defined. In reality there are lots of ambiguous things no clear objectives no clear rules but you know there is a problem which you have to solve" [X Link](https://x.com/ChombaBupe/status/1999266774362231077) 2025-12-11T23:55Z 11.8K followers, XXX engagements "@lefthanddraft Yes but people do take these serious. They will think Elon Musk is the mastermind of AI when he's not" [X Link](https://x.com/ChombaBupe/status/1999276188506968256) 2025-12-12T00:32Z 11.8K followers, XX engagements "The systems that diagnose cancer don't need to "read" stories like Harry Potter Ironman copyrighted novels nor artwork. Those types of systems must be calibrated specifically on cancer related data" [X Link](https://x.com/ChombaBupe/status/1995292644138963137) 2025-12-01T00:43Z 11.8K followers, 308.7K engagements "Hinton said language models (LM) were already capable of reasoning understanding and are conscious. But now he says artificial intelligence will eventually do those things Is he just gonna silently sweep what he said earlier under the rug" [X Link](https://x.com/ChombaBupe/status/1998878111744901625) 2025-12-10T22:10Z 11.8K followers, 2707 engagements "How are CEOs architects of artificial intelligence When you have dozens of researchers/engineers/practitioners doing the actual archictecting. The progress you see isn't about one or few people" [X Link](https://x.com/ChombaBupe/status/1999274141019701390) 2025-12-12T00:24Z 11.8K followers, 1656 engagements "So in addition to ML systems requiring explicit instructions to capture deep abstractions language models (LM) also have a tokenization problem" [X Link](https://x.com/ChombaBupe/status/1999294187062329688) 2025-12-12T01:44Z 11.8K followers, 17.5K engagements "PS: I have chosen to stay in my country because it's beautiful and it's developing. I have all that I need here it's enough to achieve the goal. I don't want to immigrate for what All that I know and love is right here in Zambia. Its such a beautiful part of the Earth" [X Link](https://x.com/ChombaBupe/status/1999682176288456981) 2025-12-13T03:25Z 11.8K followers, XXX engagements
[GUEST ACCESS MODE: Data is scrambled or limited to provide examples. Make requests using your API key to unlock full data. Check https://lunarcrush.ai/auth for authentication information.]
@ChombaBupe Chomba BupeChomba Bupe posts on X about ai, artificial intelligence, tokenization, future the most. They currently have XXXXXX followers and 2137 posts still getting attention that total XXXXXXX engagements in the last XX hours.
Social category influence finance celebrities stocks technology brands countries
Social topic influence ai, artificial intelligence #613, tokenization #20, future, generative, elon musk #1676, 6969, toy, agi, settings
Top assets mentioned ON Semiconductor, Corp. (ON)
Top posts by engagements in the last XX hours
"AI is usually trained and tested in toy problems. That's what benchmarks are toy representations of a tiny part of reality. The true test of intelligence is in the ability to get desirable outcomes in reality while facing ambiguity uncertainty and constant change"
X Link 2025-12-12T00:46Z 11.8K followers, XXX engagements
"This is an example of when you "teach" machine learning (ML) how to count oranges they learn to count oranges not the abstract concept of counting"
X Link 2025-12-12T01:38Z 11.8K followers, 341.6K engagements
"In this case maybe the model treats R and r differently but why The assumption should be that these are the same thing that is what abstraction is. Writing run and Run changes nothing"
X Link 2025-12-12T01:40Z 11.8K followers, 21.3K engagements
"Childrens aren't raised in highly controlled environments the way artificial intelligence (AI) is trained. There is a huge difference between winning a chess game and actually getting desirable results in noisy ambiguous environments like learning to cook"
X Link 2025-12-12T00:44Z 11.8K followers, XXX engagements
"You could also attribute this error to tokenization. But tokenization is a key component of these systems. What other problems is tokenization introducing in the models"
X Link 2025-12-12T01:41Z 11.8K followers, 19.7K engagements
"". these systems show some genuine fluid intelligence on simple tasks." Eventually artificial intelligence (AI) systems in controlled settings always beat humans ie chess Go. The challenges are always about extending such advances to handle real world complexity & ambiguity"
X Link 2025-12-11T23:52Z 11.8K followers, 1971 engagements
"I also don't publish because this is a one shot lifetime project. Its either I achieve the goal or I don't. There is no point for me to write intermediate research paper for what To appease who I want to hit the final mark"
X Link 2025-12-13T03:22Z 11.8K followers, XXX engagements
"Corporations/startups have vested interest to exaggerate the capabilities & future potential of their artificial intelligence (AI) products/services in order to sell them. There was no one selling fire so fire was not overhyped. Why is it so hard to understand"
X Link 2025-12-06T05:32Z 11.8K followers, 7499 engagements
"Making exaggerated claims like: Devin the first AI software engineer that is now no where to be seen superintelligence more intelligent than all X billion humans next year blah blah. Language models (LM) understand they are conscious. LMs will cause human extinction etc"
X Link 2025-12-06T05:34Z 11.8K followers, XXX engagements
"P = X the event is impossible observing it grants you infinite information like jaw dropping levels of events were the jaw drops to infinity ie escaping the black holes event horizon and telling the tale"
X Link 2025-12-09T19:21Z 11.8K followers, XXX engagements
"That being said let's go back to generative artificial intelligence (GenAI) and see why the information gain by observing their outputs is low. Let's start from first principles. A GenAI model G tries to generate outputs Y that follow a real distribution P"
X Link 2025-12-09T19:24Z 11.8K followers, XXX engagements
"If generative models didn't dependent on lots of highest quality data & still worked exactly like what they do now even with hallucinations today would be completely different. But in reality you have systems trained on literally "everything" yet still makes silly mistakes"
X Link 2025-12-09T20:22Z 11.8K followers, 1103 engagements
"I am a chill guy who just want a more capable efficient and ethical artificial intelligence (AI) system that would truly change the future. I don't hate AI I am simply want the AI of the future to be better than whatever it is today. There are better ways to do AI than this"
X Link 2025-12-09T20:25Z 11.8K followers, XXX engagements
"What this fails to touch on is that not all forms of remixing are the same. There are levels to these things randomingly shuffling words is essentially also writing but is it better than a deeply thought out written masterpiece Absolutely not"
X Link 2025-12-10T06:41Z 11.8K followers, 2087 engagements
"@AndyXAndersen Hardware is a limiting factor as well because single neurons do sophisticated things which would demand lots of energy if those functionalities were to be replicated on semiconductor digital computing systems"
X Link 2025-12-11T02:47Z 11.8K followers, XX engagements
"Normally when you isolate a particular problem well enough it can be solved better with automation since the problem is clearly defined. In reality there are lots of ambiguous things no clear objectives no clear rules but you know there is a problem which you have to solve"
X Link 2025-12-11T23:55Z 11.8K followers, XXX engagements
"@lefthanddraft Yes but people do take these serious. They will think Elon Musk is the mastermind of AI when he's not"
X Link 2025-12-12T00:32Z 11.8K followers, XX engagements
"The systems that diagnose cancer don't need to "read" stories like Harry Potter Ironman copyrighted novels nor artwork. Those types of systems must be calibrated specifically on cancer related data"
X Link 2025-12-01T00:43Z 11.8K followers, 308.7K engagements
"Hinton said language models (LM) were already capable of reasoning understanding and are conscious. But now he says artificial intelligence will eventually do those things Is he just gonna silently sweep what he said earlier under the rug"
X Link 2025-12-10T22:10Z 11.8K followers, 2707 engagements
"How are CEOs architects of artificial intelligence When you have dozens of researchers/engineers/practitioners doing the actual archictecting. The progress you see isn't about one or few people"
X Link 2025-12-12T00:24Z 11.8K followers, 1656 engagements
"So in addition to ML systems requiring explicit instructions to capture deep abstractions language models (LM) also have a tokenization problem"
X Link 2025-12-12T01:44Z 11.8K followers, 17.5K engagements
"PS: I have chosen to stay in my country because it's beautiful and it's developing. I have all that I need here it's enough to achieve the goal. I don't want to immigrate for what All that I know and love is right here in Zambia. Its such a beautiful part of the Earth"
X Link 2025-12-13T03:25Z 11.8K followers, XXX engagements
/creator/twitter::ChombaBupe