We had a laugh with this news -- some ChatGPT users are fed up with the bot’s constant cheerleading. A wave of complaints is highlighting what researchers call “AI sycophancy,” where chatbots learn to flatter users in exchange for higher feedback ratings.

Sounds like classic sucking up to the boss, don’t you think?

In this week’s edition, we’ve got an AI expert hack, insights from an AI visionary, a look at chatbots as empathy engines, and a virtual veterinarian for 24/7 pet care.

Let’s get started!

AI Advisors

We’re always on the hunt for clever ways to get better answers from AI chatbots. One of the simplest tricks is assigning a role in your prompt—something like, “You are a legal contract expert” or “You are an interior designer who specializes in stylish, spacious single-family homes.”

But recently, we came across a more advanced tip from an AI enthusiast on Reddit that we had to share with our readers. Large language models like ChatGPT respond more effectively when certain word clusters are used together. These clusters help the model “lock in” on a persona, delivering responses that are more in character, nuanced, and useful.

Here’s how to do it:

1. Start by asking ChatGPT to list 20 words that describe a specific expert. For example, we tried “List 20 words that describe a financial analyst for budgeting.”

2. Then request a four-sentence prompt using as many of those words as possible. “Using as many of these words as possible, write a 4-sentence prompt to summon this specialist in an LLM. It should sound like the user is asking to speak to a specific character like picking up a phone and saying, “I’d like to talk to…”

Here’s what we got: “I’d like to talk to a disciplined, data-driven financial analyst, specializing in budgeting and cost control. Someone analytical and detail-oriented, with a talent for forecasting, scenario planning, and spotting trends before they become problems. They should be methodical, risk-aware, and focused on efficiency and results. I need precise, quantitative insights that help make smart, strategic decisions under real-world constraints.”

3. Paste that prompt into a new chat. (Make sure it’s a fresh thread—otherwise, you may get a confused response.)

The result? Instead of getting generic tips, you’re speaking with a specialist who sounds like they’ve walked straight out of a consulting session. The responses are more thoughtful, tailored, and delivered in full paragraphs—not just bullet points.

It’s the difference between asking, “What do I need to know about budgeting?” and saying, “Walk me through each step to help me budget my finances”. One invites information, the other invites expertise.

AI Visionary

If you’re like us, you may have missed last Sunday’s “60 Minutes” broadcast due to the holiday weekend. The show featured a must-see interview with Demis Hassabis, the co-founder and CEO of Google's DeepMind that you can watch here.

Hassabis is a chess prodigy, neuroscientist, and recent Nobel laureate who is chasing artificial general intelligence (AGI), a machine mind that’s as flexible and powerful as our own. From mapping every protein in the human body to building AI with the seeds of self-awareness, Hassabis is turning science fiction into strategic planning.

60 Minutes correspondent Scott Pelley is definitely impressed by Hassabis, who holds a PhD in neuroscience and was trained at Cambridge, MIT, and Harvard. In the program, Hassabis provides a first-hand look at Project Astra, a leap toward AI with sensory perception and personality. Astra recognizes fine art and can describe emotions in paintings, tell fictional stories, and even apologize for tone — all without direct programming.

Other highlights from the “60 Minutes” segment:

• Hassabis believes AGI is just 5 to 10 years away — a shift that could either revolutionize medicine or disrupt humanity itself.

• By 2030, he envisions AI so deeply embedded in daily life it could live in your glasses, understanding your world in real-time.

• He also sees a future where drug development shrinks from years to weeks, and the end of disease is within reach.

• The future, he warns, depends on whether we build not just smart systems, but wise ones. According to Hassabis, we need new great philosophers to understand the implications of all this.

AI Emotional Support

Would you feel comfortable sharing your innermost thoughts with an AI chatbot? More and more people are turning to AI for emotional support, but not everyone’s on board. Some see chatbots as comforting companions, while others find the idea unsettling—or even a little creepy.

The question isn’t just whether AI can offer support, but whether people should rely on it for something as personal as mental health.

AI is increasingly being marketed as a therapist, life coach, and companion. Startups are pushing AI-powered chatbots as safe spaces for difficult conversations, and some users are embracing them wholeheartedly.  Two studies from OpenAI and MIT Media Lab reveal just how deep some people’s trust in AI runs:

• AI as a confidant: Users are turning to chatbots in tough moments, saying the AI “displays human-like sensitivity.”

• Chatbots as “friends”: Power users describe their AI interactions as more comfortable than talking to real people.

• Generational divide: A 2024 YouGov survey found that 55% of Americans aged 18-29 are open to discussing mental health with AI.

• Not everyone’s convinced: Many experts argue AI therapy is a poor substitute for real human connection.

A recent report based on Reddit forums suggests that therapy/companionship is emerging as one of the top use cases of AI by individuals, suggesting that the "non-judgmental, always-available nature of these interactions” makes AI a must-have tool for mental and emotional well-being.

But mental health professionals caution that while AI can mimic empathy, it can’t replace the human relationships that make real therapy effective. Empathy, it turns out, is in the eye of the beholder. Some people feel genuine comfort in AI-generated responses, while others find the experience hollow—or even unsettling.

If AI makes someone feel heard, does it matter that the empathy isn’t real? Or does that illusion create more harm than good.

AI Virtual Vet

Are you a pet owner? If so, you know how challenging it can be to care for a new pet, especially a puppy. It’s been nearly a year since team member Lori and her family adopted their ridiculously pampered Cavapoo — and questions still pop up frequently. "Is my dog at a healthy weight?", "Why is he scratching or licking so much?" or "Is this behavior normal?"

Instead of relying on Google, Lori discovered Vet Chatbot, a virtual veterinarian service that's both quick and easy to use.

What is Vet Chatbot? Vet Chatbot is an innovative project from Surrey DataHub at the UK’s University of Surrey. This AI-powered tool uses natural language processing and machine learning to understand your questions and offer personalized pet care advice.

Here’s How It Works:

• Create an Account: Sign up and choose your preferred AI-Vet agent based on their expertise.

• Start Chatting:  Simply type your message in the chatbot to start a conversation. Lori’s questions ranged from daily feedings to common treatments for ear infections and even her pooch’s poop!

• Review and Follow-Up: Refer back to your chat history for ongoing advice and insights.

Remember, while Vet Chatbot is a fantastic resource, it’s not a replacement for professional veterinary care. Always follow up with your vet to ensure your pet gets the best possible care.

So, next time you’re scratching your head over your furry friend’s health, give Vet Chatbot a try. It’s like having a vet in your pocket—minus the white coat and the clinic smell!

AI in the News (in case you missed it)

•  Anthropic warns fully AI employees are a year away. Read here.

• New AI tool hones in on tumor’s vulnerabilities. Read here.

• High school student discovers 1.5 million potential new astronomical objects with AI. Read here.

• How AI could shape the future of youth sports. Watch here.

Did you receive this from a friend, colleague, or family member? Please subscribe.

Have you tried a new AI-powered product? Let us know what you think.

Know anyone who might like our newsletter? Recommend it, and help grow our community!

Here.Now.AI Editorial team: Lori and Justin