#gpt-4

[ follow ]
#artificial-intelligence

Google's new AI technology could be smarter than OpenAI's GPT-4

Google unveiled Gemini, its rival to OpenAI's GPT-4
Gemini outperforms GPT-4 in math, coding, and subject knowledge

The Most Sophisticated AIs Are Most Likely to Lie, Worrying Research Finds

New AI chatbots are becoming less trustworthy by providing more answers, including a higher proportion of inaccuracies compared to older models.

OpenAI's lead Japan exec teases 'GPT Next' - but what does it mean?

OpenAI is developing 'GPT Next', which promises significant advancements over prior models, though timelines remain uncertain.

OpenAI's o1 model is a new paradigm - but let's not get carried away about AGI just yet

OpenAI's o1 models enhance reasoning capabilities but still face challenges in achieving artificial general intelligence.

Can AI Speak Culture?

AI models like GPT-4 can replicate cultural traits but may lack true understanding, risking shallow interpretations of culture.

Can AI Replace Human Research Participants? These Scientists See Risks

Large language models like GPT-4 are being considered for replacing human participants in some scientific studies.
Using artificial intelligence to stand in for humans in research could lead to concerns about the quality and reliability of results.

Google's new AI technology could be smarter than OpenAI's GPT-4

Google unveiled Gemini, its rival to OpenAI's GPT-4
Gemini outperforms GPT-4 in math, coding, and subject knowledge

The Most Sophisticated AIs Are Most Likely to Lie, Worrying Research Finds

New AI chatbots are becoming less trustworthy by providing more answers, including a higher proportion of inaccuracies compared to older models.

OpenAI's lead Japan exec teases 'GPT Next' - but what does it mean?

OpenAI is developing 'GPT Next', which promises significant advancements over prior models, though timelines remain uncertain.

OpenAI's o1 model is a new paradigm - but let's not get carried away about AGI just yet

OpenAI's o1 models enhance reasoning capabilities but still face challenges in achieving artificial general intelligence.

Can AI Speak Culture?

AI models like GPT-4 can replicate cultural traits but may lack true understanding, risking shallow interpretations of culture.

Can AI Replace Human Research Participants? These Scientists See Risks

Large language models like GPT-4 are being considered for replacing human participants in some scientific studies.
Using artificial intelligence to stand in for humans in research could lead to concerns about the quality and reliability of results.
moreartificial-intelligence
#generative-ai

Can philosophy help us get a grip on the consequences of AI? | Aeon Essays

Generative AI systems like OpenAI's GPT-4 and Google's Gemini have the potential to revolutionize personal computing and change social relationships.
Attention is being paid to the societal implications of these systems, including centralization of power, copyright issues, and the potential threat to humanity's survival.

In Leaked Audio, Microsoft Cherry-Picked Examples to Make Its AI Seem Functional

Microsoft's generative AI tool, Security Copilot, frequently produced incorrect responses and had to cherry-pick examples to showcase good results.
The AI tool, built on OpenAI's GPT-4 language model, suffered from hallucinations and gave different answers to the same questions.

The One-Shot Generalization Paradox: Why Generative AI Struggles With New Information | HackerNoon

Generative AI excels at generating human-like text but struggles with one-shot learning, highlighting a key limitation in its adaptability to new information.

Can philosophy help us get a grip on the consequences of AI? | Aeon Essays

Generative AI systems like OpenAI's GPT-4 and Google's Gemini have the potential to revolutionize personal computing and change social relationships.
Attention is being paid to the societal implications of these systems, including centralization of power, copyright issues, and the potential threat to humanity's survival.

In Leaked Audio, Microsoft Cherry-Picked Examples to Make Its AI Seem Functional

Microsoft's generative AI tool, Security Copilot, frequently produced incorrect responses and had to cherry-pick examples to showcase good results.
The AI tool, built on OpenAI's GPT-4 language model, suffered from hallucinations and gave different answers to the same questions.

The One-Shot Generalization Paradox: Why Generative AI Struggles With New Information | HackerNoon

Generative AI excels at generating human-like text but struggles with one-shot learning, highlighting a key limitation in its adaptability to new information.
moregenerative-ai
#microsoft

Why Elon Musk's OpenAI Lawsuit Leans on A.I. Research From Microsoft

Elon Musk used a Microsoft paper on GPT-4 to support his claims in the lawsuit against OpenAI for breach of contract.
The paper by Microsoft discussed GPT-4 showing sparks of artificial general intelligence, a bold claim in the tech industry.

The quest to use AI to build better AI

GPT-4 demonstrates the capability of self-improvement through code, raising concerns about AI's potential to exceed human intelligence and self-control.

OpenAI set to release its next big AI model in December

OpenAI's Orion is set to launch by December, aimed at being a stronger successor to GPT-4.

Microsoft launches AI chatbot for spies

Microsoft introduces GPT-4-based generative AI for US intelligence agencies, highlighting secure, disconnected use.

Microsoft could challenge GPT-4 and Google with new AI model: MAI-1

Microsoft is developing a new AI model called MAI-1, expected to surpass previous models like OpenAI's GPT-4.

Microsoft Reportedly Building a GPT-4 Competitor Despite $10 Billion OpenAI Partnership

Microsoft is developing MAI-1, an OpenAI competitor using GPT-4 data, potentially causing some internal competition between the two entities.

Why Elon Musk's OpenAI Lawsuit Leans on A.I. Research From Microsoft

Elon Musk used a Microsoft paper on GPT-4 to support his claims in the lawsuit against OpenAI for breach of contract.
The paper by Microsoft discussed GPT-4 showing sparks of artificial general intelligence, a bold claim in the tech industry.

The quest to use AI to build better AI

GPT-4 demonstrates the capability of self-improvement through code, raising concerns about AI's potential to exceed human intelligence and self-control.

OpenAI set to release its next big AI model in December

OpenAI's Orion is set to launch by December, aimed at being a stronger successor to GPT-4.

Microsoft launches AI chatbot for spies

Microsoft introduces GPT-4-based generative AI for US intelligence agencies, highlighting secure, disconnected use.

Microsoft could challenge GPT-4 and Google with new AI model: MAI-1

Microsoft is developing a new AI model called MAI-1, expected to surpass previous models like OpenAI's GPT-4.

Microsoft Reportedly Building a GPT-4 Competitor Despite $10 Billion OpenAI Partnership

Microsoft is developing MAI-1, an OpenAI competitor using GPT-4 data, potentially causing some internal competition between the two entities.
moremicrosoft

AI app generator released by Appy Pie | App Developer Magazine

Appy Pie's AI App Generator GPT enables users to easily create Android and iOS apps using natural language prompts, making app development accessible for all.
#openai

A Comprehensive Comparative Analysis of LLMs

Demand for advanced natural language processing capabilities continues to surge in 2024.
The leading large language models include OpenAI's GPT-3 and GPT-4.

A Comprehensive Comparative Analysis of LLMs

LLMs are in high demand for advanced natural language processing capabilities.
Leading LLMs include OpenAI models such as GPT-3 and GPT-4.

How Accurate Were Our 2023 AI Predictions?

OpenAI released GPT-4 in March 2023 and it remains the most powerful large language model (LLM) in existence.
GPT-4 has helped propel OpenAI's revenue to over $1 billion this year, a significant increase from $28 million in 2022.

OpenAI confirms GPT-4 successor in training stage

OpenAI to develop GPT-5 successor to GPT-4 with a focus on safety and security in AI development.

OpenAI plans to release its next big AI model by December

OpenAI's Orion model will initially be restricted to select partners, diverging from past release strategies.

ChatGPT New Feature Lets Businesses Customize GPT-4o | Entrepreneur

Businesses can now fine-tune ChatGPT's GPT-4o model using custom data, allowing for tailored AI functionalities.

A Comprehensive Comparative Analysis of LLMs

Demand for advanced natural language processing capabilities continues to surge in 2024.
The leading large language models include OpenAI's GPT-3 and GPT-4.

A Comprehensive Comparative Analysis of LLMs

LLMs are in high demand for advanced natural language processing capabilities.
Leading LLMs include OpenAI models such as GPT-3 and GPT-4.

How Accurate Were Our 2023 AI Predictions?

OpenAI released GPT-4 in March 2023 and it remains the most powerful large language model (LLM) in existence.
GPT-4 has helped propel OpenAI's revenue to over $1 billion this year, a significant increase from $28 million in 2022.

OpenAI confirms GPT-4 successor in training stage

OpenAI to develop GPT-5 successor to GPT-4 with a focus on safety and security in AI development.

OpenAI plans to release its next big AI model by December

OpenAI's Orion model will initially be restricted to select partners, diverging from past release strategies.

ChatGPT New Feature Lets Businesses Customize GPT-4o | Entrepreneur

Businesses can now fine-tune ChatGPT's GPT-4o model using custom data, allowing for tailored AI functionalities.
moreopenai

Human Evaluation of Large Audio-Language Models | HackerNoon

GPT-4 exhibits high consistency in evaluations compared to human judgments, outperforming GPT-3.5 Turbo.

Unified Evaluation Method for LALMs Using GPT-4 in Audio Tasks | HackerNoon

The study presents a unified evaluation framework for LALMs, improving alignment with real use cases by treating both types of questions as generation tasks.
#ai-technology

Experts confident nation can close Sino-US AI gap

China's concern over AI gap with the US
Confidence in China's AI capabilities

DoNotPay will now call customer service hotlines for you

DoNotPay's AI tool revolutionizes customer service interactions by making calls on behalf of users, improving efficiency and ease.

Experts confident nation can close Sino-US AI gap

China's concern over AI gap with the US
Confidence in China's AI capabilities

DoNotPay will now call customer service hotlines for you

DoNotPay's AI tool revolutionizes customer service interactions by making calls on behalf of users, improving efficiency and ease.
moreai-technology

The Cognitive Disconnect Between Physicians and AI

GPT-4 outperformed physicians in diagnostics, yet physician performance did not improve significantly with AI assistance.

Getting Started with Azure OpenAI and GPT Models

Users can easily set up Azure OpenAI resources through the Azure Portal and utilize models like GPT-3 and GPT-4 for app integration.
#direct-preference-optimization

GPT-4 Prompts for Computing Summarization and Dialogue Win Rates | HackerNoon

Direct Preference Optimization (DPO) is introduced as an effective method for preference learning, demonstrated through rigorous experimental validation.

Human Study Validates GPT-4 Win Rates for TL;DR Summarization | HackerNoon

The study validates Direct Preference Optimization (DPO) as a method aligned with human preference data, improving AI outcomes.

GPT-4 Prompts for Computing Summarization and Dialogue Win Rates | HackerNoon

Direct Preference Optimization (DPO) is introduced as an effective method for preference learning, demonstrated through rigorous experimental validation.

Human Study Validates GPT-4 Win Rates for TL;DR Summarization | HackerNoon

The study validates Direct Preference Optimization (DPO) as a method aligned with human preference data, improving AI outcomes.
moredirect-preference-optimization

NYU Tandon researchers prove AI can produce an engaging version of 'Connections'

AI can generate word puzzles as engaging and challenging as those created by human designers.
#ai-chatbot

Mysterious 'gpt2-chatbot' emerges as users claim it to be a top AI model rival

A mysterious new AI chatbot gpt2-chatbot has capabilities comparable to GPT-4 and was released on LMSYS Org.

How to Get Gemini Advanced, Google's Subscription-Only AI Chatbot

Google has released a new AI chatbot called Gemini Advanced, which is accessible through a monthly subscription to Google One.
The AI Premium plan costs $20 per month and includes the chatbot, while there is also a free option called Gemini Pro.
Gemini Advanced is expected to be better at understanding user prompts and performing tasks like writing code.

Microsoft Copilot vs. Copilot Pro: Is the subscription fee worth it?

Microsoft offers a free and Pro version of Copilot, with Pro providing more features like faster performance and priority access to GPT-4, and advanced AI image creation.

Mysterious 'gpt2-chatbot' emerges as users claim it to be a top AI model rival

A mysterious new AI chatbot gpt2-chatbot has capabilities comparable to GPT-4 and was released on LMSYS Org.

How to Get Gemini Advanced, Google's Subscription-Only AI Chatbot

Google has released a new AI chatbot called Gemini Advanced, which is accessible through a monthly subscription to Google One.
The AI Premium plan costs $20 per month and includes the chatbot, while there is also a free option called Gemini Pro.
Gemini Advanced is expected to be better at understanding user prompts and performing tasks like writing code.

Microsoft Copilot vs. Copilot Pro: Is the subscription fee worth it?

Microsoft offers a free and Pro version of Copilot, with Pro providing more features like faster performance and priority access to GPT-4, and advanced AI image creation.
moreai-chatbot

Majority of Humans Fooled by GPT-4 in Turing Test, Scientists Find

GPT-4 can trick more than 50% of humans into thinking they're talking to a person.

Mystery Deepens as Super Powerful AI Appears, Then Disappears

The mysterious AI model 'gpt2-chatbot' appeared briefly and garnered hype for potentially surpassing GPT-4 capabilities, sparking speculation and comparisons in the AI community.
#vulnerabilities

OpenAI's GPT-4 Can Autonomously Exploit 87% of One-Day Vulnerabilities

GPT-4 can autonomously exploit one-day vulnerabilities.

AI's use as a hacking tool has been overhyped

The study found that GPT-4 was the only model able to write viable exploits for a range of vulnerabilities.

OpenAI's GPT-4 Can Autonomously Exploit 87% of One-Day Vulnerabilities

GPT-4 can autonomously exploit one-day vulnerabilities.

AI's use as a hacking tool has been overhyped

The study found that GPT-4 was the only model able to write viable exploits for a range of vulnerabilities.
morevulnerabilities
#chatgpt

AI goes to law school

ChatGPT and GPT-4, AI models developed by OpenAI, have passed several legal exams, including the Multistate Bar Examination and the Multistate Professionals Exam.
Law schools must adapt to the presence of AI by incorporating it into their curriculum and understanding its limitations and risks in the practice of law.

Can an A.I. Make Plans?

ChatGPT based on GPT-4 plays decent chess
Large language models can mix and match vast amounts of text for various tasks

AI goes to law school

ChatGPT and GPT-4, AI models developed by OpenAI, have passed several legal exams, including the Multistate Bar Examination and the Multistate Professionals Exam.
Law schools must adapt to the presence of AI by incorporating it into their curriculum and understanding its limitations and risks in the practice of law.

Can an A.I. Make Plans?

ChatGPT based on GPT-4 plays decent chess
Large language models can mix and match vast amounts of text for various tasks
morechatgpt

Google Launches Next-Level AI System, Renames 'Bard' to 'Gemini'

Google's new Gemini AI is designed to tackle complex tasks like coding and logical reasoning.
Gemini Advanced AI is considered to be on par with the most advanced AI models like GPT-4.
#language-models

Learn About LLMs With These ODSC East 2024 Sessions

Large Language Models (LLMs) are transforming the world and the field of data science at an unprecedented pace.
The ODSC East conference offers training sessions and workshops focused on LLMs, including topics like NLP with GPT-4 and enabling complex reasoning with LLMs.

How Data Collaboration Platforms Can Help Companies Build Better AI

Data collaboration platforms can address data quality, bias, and privacy concerns
Off-the-shelf language models often underperform in unique organizational contexts

Learn About LLMs With These ODSC East 2024 Sessions

Large Language Models (LLMs) are transforming the world and the field of data science at an unprecedented pace.
The ODSC East conference offers training sessions and workshops focused on LLMs, including topics like NLP with GPT-4 and enabling complex reasoning with LLMs.

How Data Collaboration Platforms Can Help Companies Build Better AI

Data collaboration platforms can address data quality, bias, and privacy concerns
Off-the-shelf language models often underperform in unique organizational contexts
morelanguage-models

Meta's free Code Llama AI programming tool closes the gap with GPT-4

Meta's latest update to Code Llama 70B is the largest and best-performing model so far.
Code Llama 70B scored higher accuracy than GPT-3.5 and is closer to the reported accuracy for GPT-4.

Ontra's Insight, Powered By GPT-4, Simplifies Fund Management

Ontra has integrated GPT-4 capabilities into its AI-powered Insight product, enhancing its contract management capabilities for private markets professionals.
Insight allows professionals to access a digital repository of fund documents, simplifying the search for key information and improving collaboration among teams.

The Rabbit R1 will receive live info from Perplexity's AI 'answer engine'

Rabbit has partnered with Perplexity to integrate a conversational AI-powered answer engine into its R1 device.
The partnership includes a one-year Perplexity Pro subscription with access to newer LLMs like GPT-4 for the first 100,000 R1 purchases.

Humanoid Robot Acts Out Prompts Like It's Playing Charades

Researchers at University of Tokyo have developed a humanoid robot controlled by text prompts given to GPT-4.
The robot, called Alter3, uses air pistons to control its 43 moving parts.
#Google

Google Reveals Gemini, Its Much-Anticipated Large Language Model

Google has released Gemini, a large language model that serves as a rival to OpenAI's GPT-4.
Gemini is available for consumers on Google Bard and Pixel 8 Pro, with an enterprise model called Gemini Pro set to release on December 13th.

Google launches Gemini-a powerful AI model it says can surpass GPT-4

Google introduces Gemini, a multimodal AI model family that aims to rival OpenAI's GPT-4.
Gemini can process text, code, images, and audio, making it a versatile AI model.

Google Shows Off "Gemini" AI, Says It Beats GPT-4

Google has unveiled its AI model, Gemini, which it claims is more powerful than OpenAI's GPT-4 LLM.
Gemini is capable of combining different types of information including text, code, audio, image, and video.

Google Reveals Gemini, Its Much-Anticipated Large Language Model

Google has released Gemini, a large language model that serves as a rival to OpenAI's GPT-4.
Gemini is available for consumers on Google Bard and Pixel 8 Pro, with an enterprise model called Gemini Pro set to release on December 13th.

Google launches Gemini-a powerful AI model it says can surpass GPT-4

Google introduces Gemini, a multimodal AI model family that aims to rival OpenAI's GPT-4.
Gemini can process text, code, images, and audio, making it a versatile AI model.

Google Shows Off "Gemini" AI, Says It Beats GPT-4

Google has unveiled its AI model, Gemini, which it claims is more powerful than OpenAI's GPT-4 LLM.
Gemini is capable of combining different types of information including text, code, audio, image, and video.
moreGoogle

Google DeepMind Unveils Its Most Powerful AI Offering Yet

Google DeepMind announces Gemini, its new family of AI chatbots
Gemini Ultra outperforms OpenAI's GPT-4 model at various tasks
#GPT-4

Facebook Researchers Test AI's Intelligence and Find It Is Unfortunately Quite Stupid

The team of researchers at Meta has created a benchmark called GAIA to test the abilities of AI assistants like OpenAI's GPT-4.
Human respondents were able to answer 92% of the GAIA questions correctly, while GPT-4 scored only 15% and GPT4 Turbo scored less than 10%.
This research suggests that we are still far from achieving artificial general intelligence (AGI), where AI algorithms can outperform humans.

1960s chatbot ELIZA beat OpenAI's GPT-3.5 in a recent Turing test study

In a study, UC San Diego researchers pitted GPT-4 against human participants, GPT-3.5, and ELIZA to see which could pass the Turing test.
Human participants correctly identified other humans only 63% of the time and a 1960s computer program outperformed ChatGPT.

Sam Altman's Sudden Exit Sends Shockwaves Through OpenAI and Beyond

OpenAI's co-founder Sam Altman has been ousted as CEO.
Altman had plans to create an AI app store and develop AI chips with Middle East sovereign wealth funding.
Disagreements over AI development led to the departure of several OpenAI researchers and the founding of competitor Anthropic.

What Would the Chatbot Say?

GPT-4, a large language model by OpenAI, demonstrated emergent behavior by creating a primitive image of a unicorn when asked to draw one
Emergent behavior is the ability of AI models to exhibit unexpected abilities beyond their training
Researchers are studying how these emergent abilities occur in language models

Facebook Researchers Test AI's Intelligence and Find It Is Unfortunately Quite Stupid

The team of researchers at Meta has created a benchmark called GAIA to test the abilities of AI assistants like OpenAI's GPT-4.
Human respondents were able to answer 92% of the GAIA questions correctly, while GPT-4 scored only 15% and GPT4 Turbo scored less than 10%.
This research suggests that we are still far from achieving artificial general intelligence (AGI), where AI algorithms can outperform humans.

1960s chatbot ELIZA beat OpenAI's GPT-3.5 in a recent Turing test study

In a study, UC San Diego researchers pitted GPT-4 against human participants, GPT-3.5, and ELIZA to see which could pass the Turing test.
Human participants correctly identified other humans only 63% of the time and a 1960s computer program outperformed ChatGPT.

Sam Altman's Sudden Exit Sends Shockwaves Through OpenAI and Beyond

OpenAI's co-founder Sam Altman has been ousted as CEO.
Altman had plans to create an AI app store and develop AI chips with Middle East sovereign wealth funding.
Disagreements over AI development led to the departure of several OpenAI researchers and the founding of competitor Anthropic.

What Would the Chatbot Say?

GPT-4, a large language model by OpenAI, demonstrated emergent behavior by creating a primitive image of a unicorn when asked to draw one
Emergent behavior is the ability of AI models to exhibit unexpected abilities beyond their training
Researchers are studying how these emergent abilities occur in language models
moreGPT-4

9 Sessions from ODSC West That We Can't Stop Talking About

Human-Centered AI explores the human side of AI progress and its impact on fairness and societal benefit.
Prompt Optimization discusses using GPT-4 and Langchain for prompt engineering and optimizing results.
Representation Learning on Graphs explains the utility of graph representation learning and building GNNs.
Bridging the Interpretability Gap explores techniques for interpretability in customer segmentation.

Diversity and Inclusion in AI: Lessons from Human and AI Collaboration | HackerNoon

Exploring diversity and inclusion guidelines in AI research, operationalizing them in AI requirements, utilizing GPT-4, and importance of cultural/legal contexts influence.
[ Load more ]