MED-AI: LinkedIn, Meta, Google
This week has seen lawmakers and international institutions begin to take seriously the increasingly panicked pronouncements from AI experts about the threat the technology poses.
United Nations Secretary-General Antonio Guterres on Monday supported a proposal by some AI executives to establish an international AI watchdog body like the International Atomic Energy Agency (IAEA), saying that “alarm bells over…generative AI are deafening. And they are loudest from the developers who designed it.”
Just today the EU passed ground-breaking legislation to regulate AI with the aim of protecting public health and safety. The act classifies AI systems into four levels of risk, from minimal to unacceptable, and bans uses of AI which harm or manipulate vulnerable people (including children). Amnesty International has warned the Act may “legitimise abusive technologies” as a last minute amendment allows for a “law-enforcement exception” to remote facial recognition and biometric identification bans. Violations of the Act could mean companies are fined up to €30m or 6% of their annual global revenue.
To ensure that AI is “harnessed for the good of humanity” prime minister Rishi Sunak announced the UK government’s plans to host a global AI summit this autumn. The government said it would bring together nations, tech companies and researchers to set safety measures that “evaluate and monitor the most significant risks from AI.”
Sufficiently comforted? In that case, let’s jump into our weekly MED-AI roundup:
Google has delayed the EU launch of its AI chatbot after the Irish Data Protection Commission, the EU’s lead data protection authority, raised concerns around privacy.
The DPC deputy commissioner Graham Doyle said Google recently informed the authority of its intention to launch Bard in the EU “this week” but had not provided a detailed briefing or data protection impact assessment to the body.
Doyle commented: “[The DPC] has since sought this information as a matter of urgency and has raised a number of additional data protection questions with Google to which it awaits a response and Bard will not now launch this week.”
Meta
Last week, Meta showcased its new generative AI tech for Instagram, WhatsApp and Messenger.
Meta CEO Mark Zuckerberg announced several AI technologies including WhatsApp AI stickers and visual creation tools that would allow photo editing in Instagram Stories.
In a call with investors, Zuckerberg also teased plans to develop “chat experiences” in WhatsApp and Messenger in which users could be entertained or informed by AIs that had unique personas and skillsets.
Meta also plans to roll out internal-only products such as an AI productivity assistant and an interface for interacting with AI agents, powered by the company’s large language model LLaMA.
LinkedIn announced plans to launch its own AI-powered tool to suggest different copies of an ad.
Using data from marketers’ LinkedIn pages and Campaign manager settings (such as campaign objective, criteria and audience) the solution will offer alternative introductory text to ads.
Given it is a Microsoft company, LinkedIn will use OpenAI to power the tool which is currently in a test phase and only available to some customers in North America.
The company says it will add new functionalities and launch the feature in more languages and markets in the coming months.
Consumer wants brands to disclose their use of AI
According to an Opinium survey commissioned by the IPA, almost three-quarters of consumers believe that brands should disclose the use of AI-generated content. They also believe that fully-automated AI-driven marketing campaigns should be “carefully regulated”.
The study, which surveyed 2,000 people over the age of 18, also found that 67% of British adults think that AI “should not pretend to be a human or act as if has a personality” (can they make a Liz Truss chatbot?) and a further 75% want to be notified anytime they are not interacting with a real-person.