The next article is a guest and opinion of Johanna Rose Capeldo, founder of Data Guardians Network (D-Gn)
They are infinite data
Artificial intelligence works on data. But this data is increasingly reliable, immoral and related to legal repercussions.
The growth of obstetric intelligence is not only accelerating. He devoured everything on his way. According to what Openai, one of the expected Openai 7 billion dollars invoice In 2024 only to maintain functional models, with $ 2 billion of annual revenue. All of this was happening while Openai and Anthropic robots were chaos on web sites and raising alarm bells on the use of data on a large scale, according to his report. Business Insider.
But the problem works deeper than the costs. Artificial intelligence was built on non -transparent, old and at risk database lines. The “Data Decomposition” problem is real – models trained at the risk of unspecified, artificial or “old or” older data become less accurate over time, which leads to this Defective decision.
Legal challenges such as 12 American copyright lawsuits Against Openai and The legal problems of the anthropologist with the authors The media highlights an emerging crisis: AI is not an element by compute. The bottle is checked by Roluals of supplying trusted data.
When the synthetic is not sufficient and the scandal will not expand
Artificial data is quotations. Al -Kashti is a lawsuit awaiting its occurrence.
Artificial data has a promise of some cases of use – but not without the restaurants. He is struggling to repeat the existing differences and the depth of real situations. In health care, for example, Artificial intelligence models trained on artificial data collections can be less than performance In cases of edge, risk patient safety. In cases of high -level failure such as Google Gemini, Bias and deviant outputs It is reinforced instead of correcting it.
Meanwhile, the abstraction of the Internet is not just a public relations responsibility, but rather a structurally impasse. From the New York Times to Getty Images, lawsuits and new regulations such as AI AI are accumulated in the European Union. The notorious Tesla “Fake brand“The problem from 2022, part of which caused weak training data, explains what is happening when data sources are not identified.
While global data sizes were set to exceed 200 Zetabytes by 2025, according to Cyber security projectsMany of it is not useable or not verified. Contact and understanding are missing. Without this, confidence – and therefore, the ability to expand – is impossible.
We clearly need a new model. One where the data is created by default.
Blockchain basic data improvement
Blockchain is not only for symbols. It is the lost infrastructure of the artificial intelligence data crisis.
So, where is Blockchain suitable for this narration? How does data chaos and prevent artificial intelligence systems from nutrition to billions of data points, without approval
While the “distinctive symbol” captures headlines, it is the architecture under it that carries a real promise. The three Blockchain provides features of artificial intelligence who need desperate on the data layer: tracking or origin, verification and verification. Each of them contributes to synergy to help save artificial intelligence from legal issues, ethical challenges and data quality crises.
The tracking guarantees that each group has a verified origin. It is very similar to checking the IBM Food Trust from the farm to the shelf, we need to check a model to a source of training data. It does not ensure that no person is unable to process the record, and store important information on the series.
Finally, smart contracts automate payment flows and impose approval. In the event of a pre -determined event, verified, the smart nodes will lead to the implementation of self -programmed steps on Blockchain, without human reaction. In 2023, Lemonade has implemented a Blockchain border insurance solution For 7000 Kenyan farms. This system used smart contracts and oracles weather to release payments automatically when the pre -defined dehydration conditions are met, eliminating the need to treat manual claims.
This infrastructure fluctuates dynamic. One of the options is to use GamFied tools to name or create data. Each procedure is registered unstable. The rewards can be tracked. Approval of the series. Artificial intelligence developers are receiving ready -to -check data.
Artificial intelligence worthy of confidence requires trustworthy data
You cannot review the artificial intelligence model if you cannot review its data.
An invitation to “AI responsible” falls when the building is on invisible workers and uninterrupted sources. Antarburial lawsuits Show the real financial risks of poor data cleanliness. The general lack of confidence continues to climb, with polls that show that users Do not trust artificial intelligence models This train is on personal or unclear data.
This is no longer a legal problem anymore, it is a performance problem. MCKINSEY has shown that high -resolution data sets greatly reduce hallucinations and improve accuracy through cases of use. If we want artificial intelligence to make critical decisions in financing, health, or law, the training institution must be not viviced.
If artificial intelligence is the engine, then the data is fuel. You don’t see people who put garbage fuel in Ferrari.
New Data Economy: Why is it needed now
The distinctive symbol holds the addresses of the newspaper, but Blockchain can reinforce the entire data value chain.
We stand on the brink of economic and societal transformation. Companies they have The billions were spent on collecting data But you hardly understand its origins or risks. What we need is a new type of data economy – one based on approval, compensation and verification.
This is what it looks.
The first is the consensual collection. Users will participate in adherence, such as Briv, the privacy of the advertising ecosystem for users if they are respected and have a transparency element.
The second is fair compensation. To contribute to artificial intelligence by using their data, or the time to suspend their data, people should be compensated appropriately. Given that it is a service that individuals provide for good or unintentionally, so he took such data – which has a value that is inherent to the company – without permission or compensation that provides a difficult moral argument.
Finally, Amnesty International is responsible. With full data rates, institutions can meet compliance requirements, reduce bias and create more accurate models. This is a convincing benefit.
Forbes predicts the ability to track data on a $ 10 billion industry by 2027 – and it is not difficult to know the reason. It is the only way the balance of Amnesty International morally.
The next AI Arms will not be about who has the most graphics processing units – it will be about who has the cleanest data.
Who will build the future?
Calculating the energy and the size of the model will always be important. But real breakthroughs will not come from larger models. They will come from better foundations.
If the data, as we were told, the new oil – we will need to stop its spill, violate it, and burn it. We need to track it, appreciate it and invest in its safety.
Clean data reduces re -training courses, improves efficiency and even reduces environmental costs. Harvard research He explains that energy waste from re -trains the artificial intelligence model can compete with the emissions of small countries. Blockchain data guaranteed-can be verified from the start-makes Amnesty International smaller, faster and more green.
We can build a future where artificial intelligence creators compete not only for speed and size, but for transparency and justice.
Blockchain allows us to build artificial intelligence, and this is not only strong, but really moral. It’s time to act now – before the lawsuit, the bias scandal or hallucinations make this option for us.
