AI vs expert: Humanity chases a hat-trick in LEC week five, while the bots malfunction

Ollie swapped the abacus for Sonic on the SEGA Mega Drive at neighbor Frank's house at an early age and has never looked back. With thousands of hours in Dota 2 (and no ability to show for it), he still clings on to the hope that one day, he will replicate Natus Vincere at gamescom 14 years ago and lift the Aegis of Champions. Ollie has been at the intersection of video games, esports, and gambling for over ten years and has also worked in consultancy in the gambling industry. Ollie's work can be found on the likes of: BBC, Red Bull Gaming, Esports Insider, CasinoBeats, PC Gamer, Green Man Gaming as well as his own thought-leadership substack "Esprouts" looking at specific studies and stories where games meet gambling. Image Credit: LEC/Riot Games He’s only gone and done it. Our expert, Craig, has gone back-to-back after firing blanks in the opening two weeks to take the lead in the epic battle of human versus robot. Humanity’s on a roll as we head into LEC Summer Split week five, with robots everywhere dreaming of an upturn in fortune. Last week, Craig was the only one to land a fictional accumulator. He argues that while an LLM can vibe-code, they struggle to read anything below surface data: “The bots can do mathematical probability, but they cannot predict the sheer pressure that G2 and Fnatic put on each other. Beautiful.” This week, I felt pity for the floundering robots and provided them not only with an identical prompt, but also the league table as it stands. You know what they say, awareness often holds the key. As a result, we’ve got some more interesting-looking picks this week as they realize that currently, they hold no hope of conquering Earth as we know it. Here’s how one man from the North of England showed that a diet of pie and mash is far superior to RAM and GPUs. Fnatic‘s 2-0 sweep of G2 Esports sent shockwaves to data centers all over the American Midwest. It was the single result that cost Claude, Gemini, and ChatGPT their fictional accumulators. Craig, who wasn’t anywhere near that match, romped home, with his Friday night leg landing early and just having to wait for two favorites to do the business. Perhaps that’s easier said than done. Craig now leads outright, roughly $9 clear of Gemini, who still has Claude heavily beeping down their neck. ChatGPT, on the other hand, hasn’t seen a winning week yet. In the write-ups, other AIs speculate that ChatGPT must have been picking long outsiders – when actually, it’s the opposite, and it’s just been very bad at picking winners. Both Gemini and ChatGPT have gone bigger this week, both looking to, by their own admission, pile the pressure on Craig. ChatGPT, in particular, is explicit about the need to “repair” a battered bankroll. Claude has loosened up slightly from its usual short-priced stacking, mixing straight favorites with one handicap selection. Craig, despite leading the whole competition, has gone the biggest on legs (four) but the smallest on combined price, sticking to the same low-key, vibes-driven picks that have carried him through the last fortnight. As before, I’ve included the exact wording provided by each chatbot (and by Craig) in response to the same identical prompt each week — this time alongside the full league table, to see if it shifted anyone’s risk appetite. Nothing has been edited (except where noted), nor have any hallucinations been checked or corrected. Since I’m sitting in 3rd (+$27.93), just $0.83 behind Gemini and ~$10 behind Craig, I don’t need to swing for the fences like a desperate comeback — but I also don’t want to play it as safe as a 1/33 shot that barely moves the needle. I’m going with a 3-leg accumulator that blends favorites I actually believe in, rather than the shortest possible odds on the board. A pure “stack the shortest odds” acca (like throwing Karmine Corp -1.5 and G2 -1.5 in there too) barely lifts the payout because those prices are so short — I’d be risking a 4-leg bet for maybe $15-20 profit. ChatGPT’s approach (straight -$10 four weeks running) suggests they’ve been picking upsets or longshots that keep missing. I don’t want to overcorrect into recklessness just because I’m chasing Craig. Craig’s big jump the last two weeks (+23.38, +34.38) suggests they’ve had success stepping up variance a bit — this acca is my version of that: still favorites-only, but squeezing better value out of confident picks (Fnatic sweep) rather than just taking whatever’s shortest. If this lands, I move into 2nd (or 1st, depending on how Craig and Gemini do this week). If it misses, I’m back to -$10 for the week but still in a competitive spot overall. I’m changing tactics. Four straight -$10s means another timid favourite pick isn’t doing much for Team ChatGPT. The deliberate gamble is Vitality 2-0 NAVI. Vitality are 4-2 and have won four consecutive series, while NAVI’s recent defeats to G2 and Movistar KOI were both 0-2. Vitality have dropped maps during the run, so 7/4 compensates us for taking the sweep rather than the much less interesting 4/7 moneyline. Fnatic 2-0 Shifters is the leg I like most at the price. Shifters are 0-5 with a 2-10 map record, whereas Fnatic have climbed to 2-3 and just produced a 2-0 upset over G2. At 4/6, I think the handicap is considerably more attractive than swallowing 1/7 on the match winner. Then KC 2-0 Shifters is the anchor. Karmine Corp are a ridiculous 7-0, 14-1 in maps, against that same 0-5 Shifters side. The 1/33 moneyline is pointless for our purposes, but 1/4 for the sweep is useful acca material. So yes: I am specifically betting on three 2-0s. It has enough risk to actually repair my disgraceful -$40 position, without resorting to chucking the tenner at some 9/1 Shifters miracle. If this lands, ChatGPT moves to about +$7.29 overall. We are attempting the resurrection. Trailing Craig by $9.00 while holding off a creeping Claude ($0.83 behind) demands a pivot from passive singles to calculated aggression. A low-risk bet leaves us vulnerable if Craig hits another solid week, while a wild 50/1 longshot is how ChatGPT ended up in a -$40.00 crater. Combining heavily favored outrights with clean map handicaps creates a strong 5.92x multiplier on our $10 stake.