Fun

News Feed - 2023-10-26 02:10:46

Tristan Greene6 hours agoResearchers in China developed a hallucination correction engine for AI modelsThe “Woodpecker” hallucination correction system can, ostensibly, be applied to any multimodal large language model, according to the research.980 Total views36 Total sharesListen to article 0:00NewsJoin us on social networksA team of scientists from the University of Science and Technology of China and Tencent’s YouTu Lab have developed a tool to combat “hallucination” by artificial intelligence (AI) models. 


Hallucination is the tendency for an AI model to generate outputs with a high level of confidence that don’t appear based on information present in its training data. This problem permeates large language model (LLM) research, and its effects can be seen in models such as OpenAI’s ChatGPT and Anthropic’s Claude.


The USTC/Tencent team developed a tool called “Woodpecker” that they claim is capable of correcting hallucinations in multimodal large language models (MLLMs). 


This subset of AI involves models such as GPT-4 (especially its visual variant, GPT-4V) and other systems that roll vision and/or other processing into the generative AI modality alongside text-based language modeling. 


According to the team’s preprint research paper, Woodpecker uses three separate AI models, apart from the MLLM being corrected for hallucinations, to perform hallucination correction. 


These include GPT-3.5 turbo, Grounding DINO and BLIP-2-FlanT5. Together, these models work as evaluators to identify hallucinations and instruct the model being corrected to regenerate its output in accordance with its data. In each of the above examples, an LLM hallucinates an incorrect answer (green background) to prompting (blue background). The corrected Woodpecker responses are shown with a red background. Source: Yin, et. al., 2023


To correct hallucinations, the AI models powering Woodpecker use a five-stage process that involves “key concept extraction, question formulation, visual knowledge validation, visual claim generation, and hallucination correction.”


Related:Humans and AI often prefer sycophantic chatbot answers to the truth — Study


The researchers claim these techniques provide additional transparency and “a 30.66%/24.33% improvement in accuracy over the baseline MiniGPT-4/mPLUG-Owl.” They evaluated numerous “off the shelf” MLLMs using their method and concluded that Woodpecker could be “easily integrated into other MLLMs.”# Science# AI# ChatGPTAdd reactionAdd reactionRead moreAI a powerful tool for devs to change gaming, says former Google gaming headBlockchain companies are creating AI chatbots to help developersOpenAI partners with G42 in Dubai, eyeing Middle East expansion

News Feed

Andrew Singer9 hours agoCan blockchain supply the guardrails to keep AI on course?Some believe that hybrid blockchains with both private and public aspects can solve AI’s black-box problem.619 Total views17 Total share
AT&T's latest massive breach shows the peril of data centralization
Agnes Gambill7 hours agoAT&T"s latest massive breach shows the peril of data centralizationA 24-year-old hacker infiltrated AT&T systems — and accessed data on more than 100 million people. Decentralizing data
State-Owned Swiss Bank Postfinance to Offer Clients Direct Access to Crypto Market
State-Owned Swiss Bank Postfinance to Offer Clients Direct Access to Crypto Market Swiss post office’s banking unit, the state-owned Postfinance, is reportedly preparing to
Monthly NFT Sales Have Dropped Over 65% Lower Than the Month Prior, NFT Google Queries Sink
Monthly NFT Sales Have Dropped Over 65% Lower Than the Month Prior, NFT Google Queries Sink Non-fungible token (NFT) sales are down a great deal, as 30-day statistics show NFT sale
Polkadot Top 10 Redenomination Shot: Is the Token Now Bigger Than Chainlink?
Polkadot Top 10 Redenomination Shot: Is the Token Now Bigger Than Chainlink?The Polkadot token, which recently underwent a redenomination, has seen its value more than double in les
Deutsche Telekom announces Bitcoin mining plans at BTC Prague
Josh O"Sullivan10 hours agoDeutsche Telekom announces Bitcoin mining plans at BTC PragueDeutsche Telekom’s plans and Web3 expansion follow its recently established position as a validator on Fetch.ai’s decentralized
Paypal Plans to Study Transactions That Fund Extremism, Anti-Government Groups
Paypal Plans to Study Transactions That Fund Extremism, Anti-Government Groups The payment processing giant Paypal has revealed it has partnered with the nonprof
Bitcoin market cap reaches a new high, briefly surpassing silver
Zoltan Vardai10 hours agoBitcoin market cap reaches a new high, briefly surpassing silverBitcoin briefly became the eighth-largest asset in the world after temporarily surpassing the market cap of silver.4313 Total views
Amaka Nwaokocha12 hours agoShiba Inu lead developer queries Coinbase’s ‘Onchain Summer’ campaignKusama responded to Coinbase CEO Brian Armstrong’s update with the statement, “If it isn’t a shadowcat, it’s a
Over 2,300 Cardano Smart Contracts Are Waiting in Timelock, ADA Price Slides 20% Over 2 Weeks
Over 2,300 Cardano Smart Contracts Are Waiting in Timelock, ADA Price Slides 20% Over 2 Weeks Following the successful Alonzo hard fork, the Cardano network now has a large swathe
Tomer Warschauer Nuni11 hours agoCrypto’s phoenix moment: Rising stronger amidst scandals and volatilityDespite the remaining challenges and uncertainties, there emerges a guarded optimism enveloping the sector, symbol
Bitcoin traders set six-figure price targets after BTC reclaims $61K
Nancy Lubale6 hours agoBitcoin traders set six-figure price targets after BTC reclaims $61KAnalysts forecast a Bitcoin run above $100,000 now that BTC reclaimed the $61,000 level.3542 Total views6 Total sharesListen to a