• THE STATE OF AI
SWEDEN
About us...
“The lack of legal requirements for transparency regarding dataset sources further complicates the ethical landscape.”
$5,000 (£3,935) after the firm submitted fake citations, generated by ChatGPT, in a court filing. The AI industry proclaims that AI can’t “lie”, since it has no intention. Instead, falsehoods are called “hallucinations”. Not only does this mean that AI can make errors that are hard to spot, but bad faith actors will also have no hindrance to the use of generative AI to create whatever misleading content they wish. This means that bad faith actors can generate thousands of webpages with news articles, blogs, and science publications, all with the objective of spreading specific false information. This may also create a bias in the good faith AI that is trained online, this bias being controlled or at least affected by bad faith actors. Bad faith actors may also use generative AI to create hundreds or thousands of seemingly authentic human profiles to comment on current online events, giving the false appearance of a strong public opinion decided by a single person, faking the voice of thousands. Bias AI has varied effectiveness within specific subgroups, as biased datasets lead to inaccuracies. For example, facial recognition software may excel at recognising certain demographics, while struggling with others. Even with complete datasets, biases can persist, such as in criminal risk assessment AI trained on historical data reflecting systemic injustices. In medical applications, biased AI poses a direct harm risk, as it may not function effectively for every patient. AI in the European Union, Swedish government agencies, and public application The European Union is actively addressing concerns related to AI by developing a new AI Act. On January 22, 2024, the pre-final text of the
European Union’s Artificial Intelligence Act (“AI Act”) became accessible online. This draft was endorsed by all 27 EU Member States on 2 February and a plenary vote is scheduled for 10th – 11th April, meaning that some changes may apply. The AI Act harmonises rules for placing AI systems on the EU market, bans certain AI practices, sets requirements for high-risk systems and operation obligations, mandates transparency for AI systems intended to interact with natural persons, emotion recognition systems and biometric categorisation systems, and AI systems used to generate or manipulate image, audio or video content, and outlines market monitoring protocols. In Sweden, the Agency for Digital Government promotes AI use by establishing trust through transparency and information. While AI hasn’t been proposed to replace judges in court cases, its current applications by government agencies include translating judgments, classifying emails, transcribing speech to text, validating passports, digitising handwritten text, OCR, anonymising documents, classification of incoming e-mail, and other tasks that are suitable for AI. These applications aim to enhance efficiency without compromising public trust. Individuals or businesses should evaluate available AI services for various purposes. In legal settings, AI may assist in marketing, scheduling meetings, speech-to-text conversion, OCR for handwritten text, and generating suggestions for simple communication, such as Christmas party invitations. However, the current functionality requires thorough editing before anything generated can be used. Even this article was input into generative AI and condensed by half, the results of which you are now reading, after much editing. The full article is available on Independia Law Firm’s website, for comparison.
Independia Law Firm has accumulated extensive experience in various legal areas, successfully representing and advocating for our clients in diverse situations. We take great pride in the fact that a majority of our clients have been returning to us for many years, seeking our assistance in disputes, litigations, arbitrations, and negotiations related to working life, real estate, and business matters. Our clients return because they feel secure with our dedicated staff, who are the heart and soul of the Independia family. Our team possesses a wealth of experience and expertise in business, migration, and real estate law. Our clients can always trust that they are in capable hands, be it for family, work, or business disputes. We are proud to announce our foray into artificial intelligence, as we explore how this cutting-edge technology can revolutionise the legal industry. By embracing AI, we aim to streamline processes, enhance efficiency, and provide our clients with even more comprehensive and effective legal solutions.
hands. However, distinguishing AI-generated text from human-written text is challenging. There have been several instances in Sweden with disciplinary matters where courts must determine if students used text- generating AI to complete assignments — a very daunting task. AI-generated text is meant to sound like a plausible sentence. However, the information generated is not meant to be an accurate representation of reality. This means that modern text-generating AI will “imagine” a source of information or a factual answer to a question if it does not have the information. This culminated in a district judge in Manhattan ordering Steven Schwartz, Peter LoDuca and their law firm Levidow, Levidow & Oberman to pay
independialaw.com
“Individuals or businesses should evaluate available AI services for various purposes.”
34 | irglobal
irglobal | 35
Made with FlippingBook flipbook maker