Facebook claims its AI reduced hate by 50% despite internal documents highlighting failures

Damning reports about the ineffectiveness of Facebook’s AI in countering hate speech prompted the firm to publish a post to the contrary, but the company's own internal documents highlight serious failures.

Facebook has had a particularly rough time as of late, with a series of Wall Street Journal reports in particular claiming the company knows that “its platforms are riddled with flaws that cause harm” and “despite congressional hearings, its own pledges and numerous...

GlobalData: China is ahead of global rivals for AI ‘unicorns’

China is pulling ahead of global rivals when it comes to innovative AI “unicorns” that are pushing the technology forward.

Research from GlobalData has found that – of the 45 international AI unicorns identified – China has the largest share with 19 based in the country.

Priya Toppo, Analyst of Thematic Research at GlobalData, comments:

“China is a leading player in AI, with a number of established companies such as Baidu, Hikvision, iFlytek, Tencent,...

BenevolentAI’s drug discovery platform identifies novel target for ulcerative colitis

London-based AI pioneer BenevolentAI has identified a novel target for ulcerative colitis through its drug discovery platform.

The candidate was identified by scientists who used BenevolentAI’s target ID tools and machine learning models to identify and experimentally validate a novel biological target. Impressively, the achievement was made without any prior reference in published alliteration or patents linking the gene to ulcerative colitis.

Anne Phelan, Chief...

BCS, Chartered Institute for IT: Human reviews of AI decisions require legal protection

A leading IT industry body has warned that human reviews of AI decisions are in need of legal protection.

BCS, The Chartered Institute for IT, made the warning amid the launch of the ‘Data: A New Direction’ consultation launched by the Department for Digital, Culture, Media and Sport (DCMS).

The consultation aims to re-examine the UK’s data regulations post-Brexit. EU laws that were previously mandatory while the UK was part of the bloc – such as the...

Nvidia and Microsoft develop 530 billion parameter AI model, but it still suffers from bias

Nvidia and Microsoft have developed an incredible 530 billion parameter AI model, but it still suffers from bias.

The pair claim their Megatron-Turing Natural Language Generation (MT-NLG) model is the "most powerful monolithic transformer language model trained to date".

For comparison, OpenAI’s much-lauded GPT-3 has 175 billion parameters.

The duo trained their impressive model on 15 datasets with a total of 339 billion tokens. Various sampling weights...

Rigetti unveils ‘commercially ready’ UK-based quantum computer

Rigetti has unveiled a “commercially ready” quantum computer in the UK amid increasing local public and private investment in the technology.

The firm announced last year that it was leading a £10 million government- and industry-funded collaboration with the aim of commercialising quantum computing.

“We are excited to deliver the UK’s first quantum computer and help accelerate the development of practical algorithms and applications,” said Chad Rigetti, CEO...

DeepMind hit with class-action lawsuit over NHS health data scandal

DeepMind is facing a class-action lawsuit over its controversial use of NHS patients’ health data back in 2015.

Google-owned DeepMind was quietly given the personal records of 1.6 million patients at the Royal Free London NHS Foundation Trust.

DeepMind said that it was using the data to create a potentially life-saving app called Streams. The app was designed to alert, diagnose, and detect when patients were at risk of developing acute kidney injury. It’s currently...

Google’s MUM will start adding more context to searches

Google is getting a little help from its MUM (Multitask Unified Model) to add more context to search results.

MUM was announced at Google I/O in May and aims to transform how the web giant handles complex queries. The model uses the T5 text-to-text framework and is said to be “1,000 times more powerful” than BERT (Bidirectional Encoder Representations from Transformers), which itself was a major breakthrough when it was introduced in 2018.

One example used by Google...

Paravision boosts its computer vision and facial recognition capabilities

US-based Paravision has announced updates to boost its computer vision and facial recognition capabilities across mobile, on-premise, edge, and cloud deployments.

“From cloud to edge, Paravision’s goal is to help our partners develop and deploy transformative solutions around face recognition and computer vision,” said Joey Pritikin, Chief Product Officer at Paravision.

“With these sweeping updates to our product family, and with what has become possible in terms...

OpenAI’s latest model can summarise those tl;dr books

OpenAI has unveiled a new model that tests scalable alignment techniques by summarising those tl;dr (too long; didn’t read) books.

The model works by first summarising small sections of a book before summarising those summaries into a higher-level summary. It carries on this way – hence being a great scalable alignment test – to summarise as little or as much as desired.

You can view the complete steps on OpenAI’s website but here’s an example of where you can...