Dangerous liAIsons
The impact of artificial intelligence (AI) is being felt across most industries, with consumers able to tap the hive mind of big tech from the palm of a hand.
Read moreRead less
Being able to research on a search engine is now supercharged with AI able to review research, solve complex problems, and write and review reports with a short text or voice prompt.
However, as always, the devil is in the detail. Auto-generated ‘slop’ content fuelling dis- or mis-information, ‘deep fake’ images and videos, and spam bots have become all too apparent.
When harnessed correctly, AI can be a positive force. Look at modern cars: they’ve moved from being largely mechanical vehicles to ones driven by sensors and software. Self-driving cars have had their issues, but the technology is already helping drivers avoid collisions, assist with parking and alert emergency services in the event of a crash. AI is moving fast.
AI has immense potential but comes with responsibilities, like driving a car. You can’t buckle up and expect the tech to do all the work, but you can let it support you, with your oversight.
In the summer issue of Ethical Consumer, the publication reviewed twelve popular chatbots and no market leaders were recommended. Concerns about privacy, security, tax affairs, military links, sustainability and worker welfare abound.
However, with most software now embedding AI functionality, it’s becoming harder to avoid than ever: most new phones, TVs, cars and smart appliances have AI functions, as do many big websites.
What are the concerns?
Workers are increasingly using AI bots to help draft reports, research projects and make business decisions such as purchases, contracts and recruitment.
Beware — anything you feed into a chatbot could be used to train the models and be compromised. Personal data such as passwords, bank details, medical information, legal advice, contracts, source code, ID documents and references should all be kept out of prompts. Information given or received could be used as evidence for a breach of copyright, contract or intellectual property. It could also be used to track or profile you online. Anonymise personal data, remove any identifying detail, or replace it with made-up examples.
Which platform is safe?
Each platform has its merits. Market leader ChatGPT has a free tier that does allow you to opt out of training, but changing settings isn’t always clear. When you start a chat on the mobile app, a pop-up message tells you whether memory is on — a feature that trains the bot but is useful if you don’t want to regularly remind it of previous chats and information.
For those enabling memory, there’s also a speech-bubble icon in the top-right to have a temporary chat which won’t be recorded in the app’s memory.
AI has proved popular for settling dinner-table debates and editing photos, but remember that you should always double-check outputs, as the bots can make mistakes.
Of ChatGPT’s nearest rivals, Claude explicitly offers you the option to opt out when you register and excels at writing crisp copy for correspondence.
Google Gemini allows opt-out of sharing your data, but many tech security experts are advising users to de-Google, as the all-conquering search giant uses tech to sell ads and track you online.
Grok (embedded into X/Twitter) has concerns linked to its eccentric owner. Swiss software firm Proton specialises in privacy with encrypted tools and has its own AI tool, Lumo, promoted as confidential, with a generous free tier and paid plans from £7.99/month.