It’s not quantity it’s quality!
Researchers warn we could run out of data to train AI by 2026. What then?
We need a lot of data to train powerful, accurate and high-quality AI algorithms. For instance, ChatGPT was trained on 570 gigabytes of text data, or about 300 billion words.
Similarly, the stable diffusion algorithm (which is behind many AI image-generating apps such as DALL-E, Lensa and Midjourney) was trained on the LIAON-5B dataset comprising of 5.8 billion image-text pairs. If an algorithm is trained on an insufficient amount of data, it will produce inaccurate or low-quality outputs.
The quality of the training data is also important. Low-quality data such as social media posts or blurry photographs are easy to source, but aren’t sufficient to train high-performing AI models.
Text taken from social media platforms might be biased or prejudiced, or may include disinformation or illegal content which could be replicated by the model.
An old story.
https://abovethelaw.com/2023/11/shadow-ai-a-thorny-problem-for-law-firms/
Shadow AI: A Thorny Problem For Law Firms
There were plenty of articles written about Shadow IT – defined by Cisco as “The use of IT-related hardware or software by a department or individual without the knowledge of the IT or security group within the organization.” – Shadow IT included cloud services, software, and hardware.
Welcome to the sudden rise of Shadow AI. Its use, like that of Shadow IT, is often unknown to a law firm’s IT or security group.
… AI is everywhere, but it’s not always visible. We forget that AI is embedded in videoconferencing programs, in many legal research programs, in our e-discovery software, in the browsers we use to search for information, in our smartphones – and the list goes on and on.
Could be amusing…
Artificial intelligence experts discuss legal implications on ABA Presidential Speaker Series
“A panel of experts on artificial intelligence and how it will affect the legal landscape are featured in the next installment of the ABA Presidential Speaker Series. The program, titled “A.I. – The New Frontier,” will feature a panel of special advisers to the ABA Task Force on the Law and Artificial Intelligence. The program will be available at 3 p.m. EST on Thursday, Nov. 9. No advance registration is required. The program can be viewed here. In addition to exploring how AI has the potential to transform all aspects of society, including the practice of law, the panel will discuss the new AI Executive Order that President Biden announced on Oct. 30 — one of the first in-depth discussions by national experts examining the executive order and its ramifications.
Another kind of deepfake. Imagine my face on a two dollar bill.
https://www.androidauthority.com/google-photos-magic-editor-prohibited-edits-3383291/
Google Photos' Magic Editor will refuse to make these edits
… Summarizing the strings above, it seems Magic Editor will refuse to edit:
• Photos of ID cards, receipts, and other documents that violate Google’s GenAI terms.
• Images with personally identifiable information.
• Human faces and body parts.
• Large selections or selections that need a lot of data to be generated.
...and here I thought politicians never lied.
AI use in political campaigns raising red flags into 2024 election
… Wald said that the biggest problem that AI-generated campaign materials pose is that it promotes the concept of "the liars' dividend" where someone can claim that a fact or real-life event is a lie and a fake and sow doubt.
No comments:
Post a Comment