AI-written travel guidebooks flood the market
A New York Times investigation has revealed a barrage of travel guidebooks that appear to be compiled using Artificial Intelligence and contain sham reviews.
Books like "France Travel Guide" by Mike Steves have risen to the top of Amazon search results but contain vague, repetitive text.
The guidebooks rely on a mix of AI text generators, fake author bios and stock photos, self-publishing platforms like Amazon Kindle Direct Publishing, and the ability to solicit fake online reviews.
Well-known brands like Rick Steves, Frommer's, and Lonely Planet are still big sellers, but some worry "little bites" from the AI books are affecting sales. The faux books also often lack key safety information and have mistranslated or outdated advice that could mislead travellers.
Many of the 5-star reviews on the books also appear to be fake or nonsensical. While Amazon has policies against review manipulation, the books still maintain high ratings.
The situation in publishing underscores wider risks of AI being used to generate convincing but inaccurate or misleading content at scale.
Abrdn introduces its own AI chatbot
Global investment company and asset manager Abrdn has started onboarding employees to its in-house version of ChatGPT. The tool provides 95% complete drafts for investment reports.
Employees are encouraged to experiment with the tech to generate operational enhancements, without compromising proprietary data.
Firms are increasingly deploying AI for advising, reporting, translating and more, seeking to cut costs and complexity. Developing internal products also means they are more adapted to company use cases and offer far more security than off-the-shelf options like ChatGPT.
Following the pandemic, we’ve seen tech adoption accelerating, and investments in automation may rise as firms target manual tasks impacting margins.
Anthropic releases updated version of text-generating model, Claude Instant
Anthropic, one of OpenAI’s major competitors, has released Claude Instant 1.2, an updated version of its text-generating model.
Claude Instant 1.2 incorporates the strengths of Anthropic’s flagship model, Claude 2, and shows significant improvements in math, coding, reasoning, quote extraction, multilingual capabilities, and question answering compared to its predecessor. Anthropic claims that Claude Instant 1.2 is less prone to hallucination and more resistant to jailbreaking attempts.
Claude Instant 1.2 features a context window of 100,000 tokens (roughly the length of a novel), making it comparable to Claude 2. By comparison, OpenAI’s GPT-4 tops out at 32,000 tokens.
Anthropic’s customers and partners include platforms and tools such as Quora, DuckDuckGo, and Notion, and the startup is currently working to develop a next-gen algorithm for AI self-teaching.
The release follows a wider industry trend of releasing multiple versions of AI models to address varying market needs, as OpenAI does with its GPT-3.5 and GPT-4 models.
Generative AI skills in high demand, attracting six-figure salaries
Recent data has confirmed that generative AI skills are in high demand, attracting six-figure salaries from tech giants and non-tech companies alike. This includes Amazon and Netflix, which are paying top dollar for professionals with those skills in a bid to stay competitive.
Indeed data shows that generative AI job listings have quadrupled since the beginning of 2021, with many companies offering salaries of $100k or more to find experienced candidates.
Even entry-level positions can command six-figure salaries, with companies like Meta and Nvidia advertising jobs for generative AI research roles.
While some industry experts worry about a potential skills shortage, many companies are recognising the potential of generative AI tools to improve products and boost productivity.