Καθοδόν προς την τεχνολογική μοναδικότητα

[Yonko]

οπότε μπορω να αφηνω το pc πχ ανοιχτο ολη μερα, μαζευει credits και μετα παω και το ζουλαω σε ερωτησεις με τα credits που κερδισα. Δε θα ειναι εντελως real time, μπορει να αργει αρκετα δευτερολεπτα να σου δωσει την απαντηση, αλλα οκ τουλαχιστον δε θα κανει εκπτωσεις ποιοτητας επειδη το ρυθμισαν να κοβει το processing σε συγκεκριμενα ορια. Τα ορια θα τα βαζεις εσυ που ξοδευεις (και κερδιζεις) τα credits.

Το κακό είναι πως έτσι πρόσβαση έχουν μόνο όσοι έχουν επεξεργαστική ισχύ. Αντί για να κάνεις εσύ το "Proof-of-Work", ας το κάνει η ίδια η εταιρία, κι εσύ να την πληρώνεις σε μια άλλη μονάδα που απαιτεί παρόμοια δουλειά για να δημιουργηθεί.

Απλώς δηλαδή να πληρώνεις τα credits.

[MNP-10]

Το κακό είναι πως έτσι πρόσβαση έχουν μόνο όσοι έχουν επεξεργαστική ισχύ.

Μα τωρα σου λεει βαλε λεφτα για να εχεις προσβαση, αλλιως κανε 10 ερωτησεις τη μερα και πολυ σου ειναι. Και αυτο γιατι δεν εχουν compute power. (Και ακομα και αν βαλεις λεφτα παλι σου λεει απλα θα κανεις λιγο παραπανω ερωτησεις γιατι παλι δεν εχω compute power να σου δωσω)

Επεξεργαστες και καρτες γραφικων (igp ή καρτες) ολοι εχουν. Δλδ επεξεργαστικη ισχυ εχω (και εχουμε) να του δωσω πισω για μη κωλυεται να μου δωσει AI access...

Το compute power το σκεφτομαστε σε ορους real time αλλα εδω επειδη δεν ειναι real time, ο χρονος δουλευει υπερ μας.

Πχ καποιος εχει hardware που βγαζει 500 tflops AI... λες ουαο... που να τον κοντραρω εγω αυτον? Αλλα δε παει ετσι... γιατι εστω εγω εχω hardware που δινει 8 teraflops ΑΙ (amd apu 5700g / 2g tflop σε 32bit / 4 tflop σε 16 bit / 8 tflop σε 8 bit). E απλα εμενα για καθε δευτερολεπτο του αλλου, θελει 1 λεπτο να μαζεψει compute. Αλλα θα το μαζεψει. Αν το αφησω κατι ωρες / μερες θα εχω το equivalent οχι του αλλου που εχει 500 tflop AI αλλα πολλαπλασιο κατα πολυ. Και τοτε μπορω να το ξοδεψω με μια μεγαλη συζητηση με complex requests που θα γινονται distribute στο δικτυο για να το κανουν process χιλιαδες αλλα ατομα.

Με συμφερει μακραν γιατι αφενος δε χρειαζεται hardware, αφετερου δεν χρειαζεται συνδρομη και τελος εχω τσαμπα ρευμα (solar). Βεβαια δεν εχω αρχισει να υπολογιζω ρευματα ουτε στον αλλο (σεναριο κεντρικου σερβερ με nvidia ή AI accels), ουτε σε μενα... γιατι και οι 2 καταναλωνουν.

[Φιλόσοφος_Στ@ρχίδας]

Το ChatGPT είναι biased.
To Gemini είναι biased.
Το Llama είναι biased.
Ο Musk δεν μπορεί να φτιάξει anti-woke LLM και επειδή δεν τα κατάφερε, το έκανε open source.

Μάλλον δεν είναι στραβός ο γιαλός, μάλλον εσείς αρμενίζετε στραβά

[flamelab]

Scarlett Johansson ‘Angered’ By ChatGPT Voice That Sounded ‘Eerily’ Like Her

[dimyok]

"different actor" our ass

[acct]

OpenAI: A landmark multi-year global partnership with News Corp

In addition to providing content, News Corp will share journalistic expertise to help ensure the highest journalism standards are present across OpenAI’s offering.

Την ίδια στιγμή, η κατάσταση στην OpenAI είναι αυτή:
Gretchen Krueger @GretchenMarina

I resigned a few hours before hearing the news about @ilyasut and @janleike, and I made my decision independently. I share their concerns. I also have additional and overlapping concerns.

Καλά πάει.

[Φιλόσοφος_Στ@ρχίδας]

Και μάλιστα ενώ τρέχουν πόσες μηνύσεις από εφημερίδες. 250 εκ μου φαίνονται λίγα πάντως.

[acct]

ενώ τρέχουν πόσες μηνύσεις από εφημερίδες

...και όχι μόνο.

Leaked OpenAI documents reveal aggressive tactics toward former employees

Throughout the hundreds of pages of documents leaked to Vox, a pattern emerges. Getting ex-employees to sign the ultra-restrictive nondisparagement and nondisclosure agreement involved threatening to cancel their equity — but it also involved much more.
[...]
The first is when I reviewed an employee termination contract and saw it casually stating that as “consideration” for signing this super-strict agreement, the employee would get to keep their already vested equity. That might not mean much to people outside the tech world, but I knew that it meant OpenAI had crossed a line many in tech consider close to sacred.

The second “woah” moment was when I reviewed the second termination agreement sent to one ex-employee who'd challenged the legality of OpenAI's scheme. The company, rather than defending the legality of its approach, had just jumped ship to a new approach.

That led to the third “woah” moment. I read through the incorporation document that the company cited as the reason it had the authority to do this and confirmed that it did seem to give the company a lot of license to take back vested equity and block employees from selling it. So I scrolled down to the signature page, wondering who at OpenAI had set all this up. The page had three signatures. All three of them were Sam Altman.

Και παρ' όλα αυτά, κομβικοί άνθρωποι παραιτούνται και αρνούνται να υπογράψουν, με σχόλια όπως παρακάτω:

[MNP-10]

Tell me the 10 biggest false assumption that humanity (as a consensus) has, per your own view.

ChatGPT
Identifying the biggest false assumptions humanity holds requires examining deeply ingrained beliefs and societal norms. Here are ten significant false assumptions, in my view, that often guide human behavior and decision-making:

1. Infinite Economic Growth is Sustainable

Assumption: Economic growth can continue indefinitely without negative consequences.
Reality: The planet has finite resources, and continuous growth leads to environmental degradation, resource depletion, and unsustainable consumption patterns.

2. Technology Can Solve All Problems
Assumption: Technological advancements will eventually solve all of humanity’s problems.
Reality: While technology can address many issues, it cannot solve problems rooted in human behavior, ethics, and socio-economic structures. Overreliance on technology can also create new problems.

3. Humans are Separate from Nature
Assumption: Humans are distinct from and superior to nature.
Reality: Humans are an integral part of the natural ecosystem. This separation mindset leads to environmental exploitation and degradation, ignoring the interdependence of all life forms.

4. Material Wealth Equals Happiness
Assumption: Accumulating material wealth and possessions leads to happiness and fulfillment.
Reality: Beyond a certain point, material wealth does not significantly increase happiness. Emotional well-being, relationships, and a sense of purpose are more critical to long-term happiness.

5. More is Always Better
Assumption: More consumption, more production, and more growth are always beneficial.
Reality: Excessive consumption and production often lead to waste, environmental harm, and diminishing returns in terms of well-being and quality of life.

6. Humans are Rational Actors
Assumption: People make decisions based purely on rational self-interest and logical analysis.
Reality: Human behavior is often influenced by emotions, biases, social pressures, and irrational factors, leading to decisions that are not always logical or in one's best interest.

7. Science and Technology are Value-Neutral
Assumption: Scientific discoveries and technological advancements are inherently neutral and only their application determines their moral value.
Reality: The development and application of science and technology are influenced by human values, ethics, and socio-political contexts. They can perpetuate inequalities and cause harm if not guided by ethical considerations.

8. The Market Will Self-Correct
Assumption: Free markets will naturally correct themselves and lead to optimal outcomes for society.
Reality: Markets can fail, leading to economic disparities, environmental damage, and social injustices. Regulation and intervention are often necessary to address market failures and ensure equitable outcomes.

9. Historical Progress is Linear
Assumption: Human history is a steady progression toward betterment and enlightenment.
Reality: History is marked by cycles, regressions, and discontinuities. Progress is not guaranteed, and societies can and do regress under various pressures.

10. Humanity is the Pinnacle of Evolution
Assumption: Humans are the final and most advanced product of evolution.
Reality: Evolution is an ongoing process with no predetermined endpoint. Viewing humanity as the pinnacle can foster hubris and neglect the importance of biodiversity and the contributions of other species to the planet's health.

Challenging these assumptions requires critical thinking, humility, and a willingness to embrace more sustainable and holistic approaches to living and problem-solving.

.........

[Yonko]

Τι θέλεις να σου πει; Αφού είναι Democrat.

[badweed]

ουσιαστικα δεν θελει και πολυ εμπειρια απο την ζωη για να καταληξεις σε αυτα που αναφερει το gpt4 στο πονημα του για τα 10 σημεια αυταπατης του ανθρωπου .

νομιζω και στην αρχαιοτητα τα συνανταμε σαν τα θανασιμα αμαρτηματα ,τα παθη , οτι πρεσβευει η μοντερνα κατηχηση ως δρομο για το μελλον . τον χαμο .

Off Topic

παω να δω το se7en

[MNP-10]

Ειναι βεβαιο οτι η ανθρωποτητα εχει biases και υποθεσεις οι οποιες δεν ειναι απαραιτητο οτι ισχυουν, οπότε με το ΑΙ αυτα θα φανουν.... σε 2-3 versions θα μας κανει σουρωτηρι αυτο που εμεις θεωρουμε ως "λογικο"... εντωμεταξυ σε καποια θεματα κανει inject common sense που απουσιαζει...

πχ σε παρομοιο prompt

in order to effectively find solutions you must have proper definitions of a problem, proper data and proper analysis of data, without assumptions, logic leaps, failed logic or consensual agreements that are made without hard basis... so identify areas where humanity is trying to solve problems that don't even exist or are not even problems, per a more enlightened view / your view, or where the fundamentals in which humanity starts to approach the problem are wrong altogether. Humanity definitely needs a more neutral look / an outside look that can give it a better perspective and balance.

...μερικες απ'τις απαντησεις:

Geoengineering: Large-scale geoengineering projects, like solar radiation management, come with unpredictable risks and might divert attention from more sustainable and immediate actions like reducing carbon emissions and conserving ecosystems.
...
Overdiagnosis of Medical Conditions: The medical community sometimes identifies and treats conditions that might never cause symptoms or lead to significant health problems, leading to unnecessary treatments and anxiety (e.g., certain types of cancers that are very slow-growing).
...
Infinite Growth Model: This model assumes that infinite growth is possible on a planet with finite resources, leading to environmental degradation and unsustainable practices. The focus should shift towards sustainable development and well-being metrics rather than GDP alone.
...
Mass Surveillance: While aimed at improving security, mass surveillance can lead to significant privacy issues and might not be as effective as targeted, intelligence-led policing. The trade-offs between security and personal freedom need careful consideration.
...
Monoculture Farming: This practice aims to maximize crop yields but leads to soil depletion, loss of biodiversity, and increased vulnerability to pests and diseases. A more holistic approach, like permaculture or agroecology, might be more sustainable.
...
AI for All Problems: There's a strong push to apply AI solutions to every conceivable problem, sometimes without considering whether simpler, more human-centered solutions might be more effective. Over-reliance on AI can lead to job displacement, ethical concerns, and loss of human skills and judgment.
...
Focus on Mars and Beyond: While space exploration and potential colonization can inspire and drive technological advances, the notion that we can solve Earth's problems by moving to another planet is flawed. Issues like climate change, resource depletion, and environmental degradation need solutions here and now, on Earth.
...
Criminalization over Treatment: The global war on drugs focuses heavily on criminalization rather than treating substance abuse as a public health issue. This approach has led to overpopulated prisons, increased crime rates, and significant social costs, without effectively reducing drug use or addressing the underlying causes of addiction.
...
Standardized Testing: Overemphasis on standardized testing can stifle creativity, critical thinking, and the love of learning. Education systems that prioritize test scores over holistic development often fail to prepare students for real-world challenges and opportunities.
...
Overmedication: The reliance on pharmaceutical solutions for mental health issues can sometimes overshadow the benefits of therapy, lifestyle changes, and community support. Addressing mental health holistically requires a balance of medication, counseling, social connection, and healthy living.
...
Focus on Disease Treatment over Prevention: Public health policies often prioritize treatment over prevention. Investing in preventive measures such as vaccination, education, and healthy living initiatives can be more cost-effective and impactful in the long run.
...
Superficial CSR: Many corporate social responsibility (CSR) initiatives are more about image management than genuine impact. (επιτελους καποιος το πε ) True CSR should involve fundamental changes in business practices, focusing on sustainability, ethical labor practices, and community engagement.

Το προβλημα ειναι οτι οταν ξεκιναει και το ιδιο με training biases δε μπορει να ειναι 100% αντικειμενικο... πχ αναφορα σε CO2 problem...

Αν το παει ουδετερα (χωρις σοβαρο training bias) θα γινει jet στο μελλον.

[MNP-10]

https://www.zerohedge.com/technology...eningly-broken

Meanwhile

https://x.ai/blog/series-b

Our Series B funding round of $6 billion with participation from key investors including Valor Equity Partners, Vy Capital, Andreessen Horowitz, Sequoia Capital, Fidelity Management & Research Company, Prince Alwaleed Bin Talal and Kingdom Holding, amongst others.

xAI has made significant strides over the past year. From the announcement of the company in July 2023, to the release of Grok-1 on X in November, to the recent announcements of the improved Grok-1.5 model with long context capability, to Grok-1.5V with image understanding, xAI’s model capabilities have improved rapidly. With the open-source release of Grok-1, xAI has opened doors for advancements in various applications, optimizations, and extensions of the model.

xAI will continue on this steep trajectory of progress over the coming months, with multiple exciting technology updates and products soon to be announced. The funds from the round will be used to take xAI’s first products to market, build advanced infrastructure, and accelerate the research and development of future technologies.

xAI is primarily focused on the development of advanced AI systems that are truthful, competent, and maximally beneficial for all of humanity. The company’s mission is to understand the true nature of the universe.